alexei-led
449 verified skills12,715 total stars
configuring-git-hygiene
Configure safe git workflow hygiene: pre-commit/pre-push hooks, Gitleaks secret scanning, .gitignore rules, local git config, and guardrails. Use when setting up git hooks, gitleaks/git leaks, staged pre-commit checks, pre-push validation, core.hooksPath, .gitignore, or git config best practices. NOT for creating commits (use committing-code), cleaning branches/worktrees (use cleanup-git), or creating worktrees (use using-git-worktrees).
development33
researching-web
Web research via platform web tools. Use for technical comparisons, recent facts, ecosystem news, best practices, standards, or questions needing grounded web evidence. NOT for API syntax lookup or code examples — use looking-up-docs for those. NOT for repo-specific questions — search local files first.
tools33
evolving-config
Audit and improve AI coding-agent configuration. Use when reviewing or changing Claude Code, Pi, Codex, skill, agent, hook, MCP, permission, package, or generated-export setup. Default is review-only; fixes require explicit user approval or --fix. NOT for application config, git hygiene, code bugs, ordinary docs, or generated files without their source.
tools33
cleanup-git
Remove merged local branches and stale git worktrees. Use when the user says "cleanup branches", "prune worktrees", "tidy git", "remove merged branches", "delete merged branches", "gone branches", or wants to clean local git state. NOT for creating commits, creating worktrees, or configuring git hooks.
development33
configuring-git-hygiene
Configure safe git workflow hygiene: pre-commit/pre-push hooks, Gitleaks secret scanning, .gitignore rules, local git config, and guardrails. Use when setting up git hooks, gitleaks/git leaks, staged pre-commit checks, pre-push validation, core.hooksPath, .gitignore, or git config best practices. NOT for creating commits (use committing-code), cleaning branches/worktrees (use cleanup-git), or creating worktrees (use using-git-worktrees).
development33
reviewing-instructions
Use when asked to lint, audit, review, or score AI-facing instruction files such as SKILL.md, AGENT.md, AGENTS.md, CLAUDE.md, platform body.md files, prompt files, rules, policies, and agent-facing references. NOT for application code review, harness configuration review, ordinary docs, tests, or generated build output.
development33
using-git-worktrees
Creates isolated git worktrees for parallel development. Use when starting feature work needing isolation or working on multiple branches simultaneously. NOT for simple branch switching, bulk branch cleanup (use cleanup-git), or git hook/config setup (use configuring-git-hygiene).
development33
writing-shell
Idiomatic shell development for POSIX sh, Bash, Zsh, Fish, hooks, CI shell steps, and scriptable CLI glue. Use when writing or changing `.sh`, `.bash`, `.zsh`, `.fish`, `.bats`, shell functions, shell pipelines, or command-runner recipes. Emphasizes portability, quoting, safe filesystem/process handling, non-TUI CLI tools, ShellCheck, shfmt, Bats, and ShellSpec. NOT for Python, TypeScript, Go, web code, or infrastructure operations.
tools33
writing-typescript
Idiomatic TypeScript development. Use when writing TypeScript code, Node.js services, React apps, or TypeScript design advice. Emphasizes strict typing, boundary validation, composition, behavior tests, and project-configured tooling. NOT for Go, Python, plain HTML/CSS/JS, or server-rendered templates.
tools33
writing-web
Simple web development with HTML, CSS, JS, and HTMX. Use when working with .html, .css, or .htmx files, web templates, stylesheets, or vanilla JS scripts. NOT for React/Vue/Angular (use writing-typescript) or Node.js backends.
development33
brainstorming-ideas
Brainstorm ideas and stress-test draft plans before coding. Use when brainstorming, exploring approaches, designing a feature/API/flow, grilling or debating a bounded plan, challenging assumptions, or resolving design-blocking terminology. NOT for implementation task breakdown. NOT for generic technology comparisons or best-practice research; use researching-web. NOT for docs updates; use documenting-code.
development33
browser-automation
Browser automation for rendered UI exploration, validation, screenshots, recordings, and end-to-end flows. Use when a task needs an actual browser or rendered DOM: inspect UI state, click/fill forms, debug frontend behavior, capture evidence, verify a feature, or run/generate browser tests. NOT for API checks or pure logic tests where curl, unit tests, or JSDOM is cheaper.
tools33
spec-flow
Use when planning, executing, checkpointing, finishing, or inspecting lightweight spec-driven work. Runs one task at a time using `.spec/` markdown files and the bundled `specctl` helper. NOT for broad product discovery beyond a short requirement interview.
tools33
committing-code
Create normal git commits with logical grouping. Use when committing, saving changes, creating commits, or grouping work into commits. NOT for amending, rebasing, force-pushing, or rewriting history.
development33
brainstorming-ideas
Brainstorm ideas and stress-test draft plans before coding. Use when brainstorming, exploring approaches, designing a feature/API/flow, grilling or debating a bounded plan, challenging assumptions, or resolving design-blocking terminology. NOT for implementation task breakdown. NOT for generic technology comparisons or best-practice research; use researching-web. NOT for docs updates; use documenting-code.
development33
configuring-git-hygiene
Configure safe git workflow hygiene: pre-commit/pre-push hooks, Gitleaks secret scanning, .gitignore rules, local git config, and guardrails. Use when setting up git hooks, gitleaks/git leaks, staged pre-commit checks, pre-push validation, core.hooksPath, .gitignore, or git config best practices. NOT for creating commits (use committing-code), cleaning branches/worktrees (use cleanup-git), or creating worktrees (use using-git-worktrees).
development33
writing-go
Idiomatic Go development. Use when writing Go code, designing APIs, reviewing Go implementations, or changing Go tests. Follow the module's target Go version. Prefer stdlib, concrete types, explicit errors, context propagation, and behavior tests. NOT for Python, TypeScript, shell scripts, or infra-only work.
development33
operating-infra
Author, inspect, troubleshoot, and review infrastructure across IaC, Kubernetes, cloud resources, containers, CI/CD, and Linux hosts. Use when changing Terraform/OpenTofu, Kubernetes, Helm, Kustomize, Dockerfiles, GitHub Actions, AWS, GCP, Cloud Run, BigQuery, IAM, logs, instances, or service health. NOT for deploy/apply/rollback workflows (see deploying-infra). NOT for shell scripts or generic command pipelines (see writing-shell).
testing33
writing-python
Idiomatic Python 3.12+ development. Use when writing Python code, CLI tools, scripts, or services. Emphasizes stdlib, type hints, uv/ruff/pyright toolchain, and minimal dependencies. NOT for Go, TypeScript, or shell-only tasks.
tools33
refactoring-code
Batch behavior-preserving refactors for multi-file, repeated-pattern, large-file, rename, move, extract, split, or restructure work. Use for "refactor across files", "batch rename", "update pattern everywhere", large files (500+ lines), or 5+ coordinated edits in one file. NOT for single targeted edits, behavior changes or bug fixes (use fixing-code), test-only refactors (use improving-tests), code review (use reviewing-code), or architecture redesign (use architecture-design/review).
development33
looking-up-docs
Find current, version-correct library/API/framework docs through one lookup workflow. Use when the user says "look up docs", "how to use", "API for", "syntax for", "examples of", "show me the docs", mentions "ctx7"/"Context7", passes a `/org/project` library ID, or wants the latest/current/actual behavior of a library, framework, CLI, or API. NOT for comparisons, best-practice surveys, or recent ecosystem news — use researching-web.
tools33
using-git-worktrees
Creates isolated git worktrees for parallel development. Use when starting feature work needing isolation or working on multiple branches simultaneously. NOT for simple branch switching, bulk branch cleanup (use cleanup-git), or git hook/config setup (use configuring-git-hygiene).
development33
deploying-infra
Validate infrastructure changes and, after explicit confirmation, apply Terraform, Helm, Kustomize, or Kubernetes deployments. Use when the user says "deploy", "deploy to staging", "terraform apply", "helm upgrade", "kubectl apply", "rollout", "deploy check", "validate deployment", or "validate infrastructure". Dockerfiles and GitHub Actions are validate-only here. NOT for ongoing service troubleshooting, cloud inspection, rollback investigation, or authoring infra from scratch; use operating-infra for those.
development33
playwright-skill
Support-only Playwright runtime/reference for browser-automation — dev-server detection, a Node.js script runner, quiet screenshot helpers, SPA readiness helpers, and custom HTTP headers. Use when browser-automation selects the bundled Playwright fallback; do not route user intent here directly.
tools33
committing-code
Create normal git commits with logical grouping. Use when committing, saving changes, creating commits, or grouping work into commits. NOT for amending, rebasing, force-pushing, or rewriting history.
development33
improving-tests
Improve test design and coverage with behavior-focused tests, useful seams, characterization tests, TDD, and test refactoring. Use when improving tests, adding coverage, refactoring brittle tests, removing test waste, or working test-first. NOT for fixing production bugs (use fixing-code), production-code refactors (use refactoring-code), or reviewing non-test code quality (use reviewing-code).
development33
fixing-code
Fix code defects with a reproducible feedback loop, root-cause diagnosis, minimal patch, regression test, and clean verification. Use when debugging, diagnosing, or resolving lint/test/build failures. NOT for behavior-preserving refactors (use refactoring-code), test-suite cleanup without a production bug (use improving-tests), or code review findings without fixes (use reviewing-code).
development33
cleanup-git
Remove merged local branches and stale git worktrees. Use when the user says "cleanup branches", "prune worktrees", "tidy git", "remove merged branches", "delete merged branches", "gone branches", or wants to clean local git state. NOT for creating commits, creating worktrees, or configuring git hooks.
development33
evolving-config
Audit and improve AI coding-agent configuration. Use when reviewing or changing Claude Code, Pi, Codex, skill, agent, hook, MCP, permission, package, or generated-export setup. Default is review-only; fixes require explicit user approval or --fix. NOT for application config, git hygiene, code bugs, ordinary docs, or generated files without their source.
tools33
documenting-code
Create or update human-facing docs, agent-facing instructions, architecture docs, API docs, README content, and useful code comments from implementation facts. Use when docs are stale, missing, or must reflect code changes. NOT for code-quality review, prompt scoring, speculative docs, or ADRs unless explicitly requested.
development33
reviewing-code
Use when reviewing changed code, PRs, diffs, or specific files. Finds evidence-backed defects in security, correctness, tests, reliability, performance, maintainability, and docs. Supports quick, standard, deep, team, and external-review modes. NOT for repo-wide architecture review, general codebase exploration, fixing issues (use fixing-code), improving tests without a code review (use improving-tests), or applying refactors (use refactoring-code).
development33
brainstorming-ideas
Brainstorm ideas and stress-test draft plans before coding. Use when brainstorming, exploring approaches, designing a feature/API/flow, grilling or debating a bounded plan, challenging assumptions, or resolving design-blocking terminology. NOT for implementation task breakdown. NOT for generic technology comparisons or best-practice research; use researching-web. NOT for docs updates; use documenting-code.
development33
playwright-skill
Support-only Playwright runtime/reference for browser-automation — dev-server detection, a Node.js script runner, quiet screenshot helpers, SPA readiness helpers, and custom HTTP headers. Use when browser-automation selects the bundled Playwright fallback; do not route user intent here directly.
tools33
researching-web
Web research via platform web tools. Use for technical comparisons, recent facts, ecosystem news, best practices, standards, or questions needing grounded web evidence. NOT for API syntax lookup or code examples — use looking-up-docs for those. NOT for repo-specific questions — search local files first.
tools33
reviewing-instructions
Use when asked to lint, audit, review, or score AI-facing instruction files such as SKILL.md, AGENT.md, AGENTS.md, CLAUDE.md, platform body.md files, prompt files, rules, policies, and agent-facing references. NOT for application code review, harness configuration review, ordinary docs, tests, or generated build output.
development33
writing-go
Idiomatic Go development. Use when writing Go code, designing APIs, reviewing Go implementations, or changing Go tests. Follow the module's target Go version. Prefer stdlib, concrete types, explicit errors, context propagation, and behavior tests. NOT for Python, TypeScript, shell scripts, or infra-only work.
development33
writing-python
Idiomatic Python 3.12+ development. Use when writing Python code, CLI tools, scripts, or services. Emphasizes stdlib, type hints, uv/ruff/pyright toolchain, and minimal dependencies. NOT for Go, TypeScript, or shell-only tasks.
tools33
operating-infra
Author, inspect, troubleshoot, and review infrastructure across IaC, Kubernetes, cloud resources, containers, CI/CD, and Linux hosts. Use when changing Terraform/OpenTofu, Kubernetes, Helm, Kustomize, Dockerfiles, GitHub Actions, AWS, GCP, Cloud Run, BigQuery, IAM, logs, instances, or service health. NOT for deploy/apply/rollback workflows (see deploying-infra). NOT for shell scripts or generic command pipelines (see writing-shell).
testing33
writing-shell
Idiomatic shell development for POSIX sh, Bash, Zsh, Fish, hooks, CI shell steps, and scriptable CLI glue. Use when writing or changing `.sh`, `.bash`, `.zsh`, `.fish`, `.bats`, shell functions, shell pipelines, or command-runner recipes. Emphasizes portability, quoting, safe filesystem/process handling, non-TUI CLI tools, ShellCheck, shfmt, Bats, and ShellSpec. NOT for Python, TypeScript, Go, web code, or infrastructure operations.
tools33
researching-web
Web research via platform web tools. Use for technical comparisons, recent facts, ecosystem news, best practices, standards, or questions needing grounded web evidence. NOT for API syntax lookup or code examples — use looking-up-docs for those. NOT for repo-specific questions — search local files first.
tools33
writing-web
Simple web development with HTML, CSS, JS, and HTMX. Use when working with .html, .css, or .htmx files, web templates, stylesheets, or vanilla JS scripts. NOT for React/Vue/Angular (use writing-typescript) or Node.js backends.
development33
writing-typescript
Idiomatic TypeScript development. Use when writing TypeScript code, Node.js services, React apps, or TypeScript design advice. Emphasizes strict typing, boundary validation, composition, behavior tests, and project-configured tooling. NOT for Go, Python, plain HTML/CSS/JS, or server-rendered templates.
tools33
browser-automation
Browser automation for rendered UI exploration, validation, screenshots, recordings, and end-to-end flows. Use when a task needs an actual browser or rendered DOM: inspect UI state, click/fill forms, debug frontend behavior, capture evidence, verify a feature, or run/generate browser tests. NOT for API checks or pure logic tests where curl, unit tests, or JSDOM is cheaper.
tools33
spec-flow
Use when planning, executing, checkpointing, finishing, or inspecting lightweight spec-driven work. Runs one task at a time using `.spec/` markdown files and the bundled `specctl` helper. NOT for broad product discovery beyond a short requirement interview.
tools33
documenting-code
Create or update human-facing docs, agent-facing instructions, architecture docs, API docs, README content, and useful code comments from implementation facts. Use when docs are stale, missing, or must reflect code changes. NOT for code-quality review, prompt scoring, speculative docs, or ADRs unless explicitly requested.
development33
committing-code
Create normal git commits with logical grouping. Use when committing, saving changes, creating commits, or grouping work into commits. NOT for amending, rebasing, force-pushing, or rewriting history.
development33
documenting-code
Create or update human-facing docs, agent-facing instructions, architecture docs, API docs, README content, and useful code comments from implementation facts. Use when docs are stale, missing, or must reflect code changes. NOT for code-quality review, prompt scoring, speculative docs, or ADRs unless explicitly requested.
development33
fixing-code
Fix code defects with a reproducible feedback loop, root-cause diagnosis, minimal patch, regression test, and clean verification. Use when debugging, diagnosing, or resolving lint/test/build failures. NOT for behavior-preserving refactors (use refactoring-code), test-suite cleanup without a production bug (use improving-tests), or code review findings without fixes (use reviewing-code).
development33
refactoring-code
Batch behavior-preserving refactors for multi-file, repeated-pattern, large-file, rename, move, extract, split, or restructure work. Use for "refactor across files", "batch rename", "update pattern everywhere", large files (500+ lines), or 5+ coordinated edits in one file. NOT for single targeted edits, behavior changes or bug fixes (use fixing-code), test-only refactors (use improving-tests), code review (use reviewing-code), or architecture redesign (use architecture-design/review).
development33
fixing-code
Fix code defects with a reproducible feedback loop, root-cause diagnosis, minimal patch, regression test, and clean verification. Use when debugging, diagnosing, or resolving lint/test/build failures. NOT for behavior-preserving refactors (use refactoring-code), test-suite cleanup without a production bug (use improving-tests), or code review findings without fixes (use reviewing-code).
development33
evolving-config
Audit and improve AI coding-agent configuration. Use when reviewing or changing Claude Code, Pi, Codex, skill, agent, hook, MCP, permission, package, or generated-export setup. Default is review-only; fixes require explicit user approval or --fix. NOT for application config, git hygiene, code bugs, ordinary docs, or generated files without their source.
tools33
improving-tests
Improve test design and coverage with behavior-focused tests, useful seams, characterization tests, TDD, and test refactoring. Use when improving tests, adding coverage, refactoring brittle tests, removing test waste, or working test-first. NOT for fixing production bugs (use fixing-code), production-code refactors (use refactoring-code), or reviewing non-test code quality (use reviewing-code).
development33
researching-web
Web research via platform web tools. Use for technical comparisons, recent facts, ecosystem news, best practices, standards, or questions needing grounded web evidence. NOT for API syntax lookup or code examples — use looking-up-docs for those. NOT for repo-specific questions — search local files first.
tools33
reviewing-code
Use when reviewing changed code, PRs, diffs, or specific files. Finds evidence-backed defects in security, correctness, tests, reliability, performance, maintainability, and docs. Supports quick, standard, deep, team, and external-review modes. NOT for repo-wide architecture review, general codebase exploration, fixing issues (use fixing-code), improving tests without a code review (use improving-tests), or applying refactors (use refactoring-code).
development33
cleanup-git
Remove merged local branches and stale git worktrees. Use when the user says "cleanup branches", "prune worktrees", "tidy git", "remove merged branches", "delete merged branches", "gone branches", or wants to clean local git state. NOT for creating commits, creating worktrees, or configuring git hooks.
development33
brainstorming-ideas
Brainstorm ideas and stress-test draft plans before coding. Use when brainstorming, exploring approaches, designing a feature/API/flow, grilling or debating a bounded plan, challenging assumptions, or resolving design-blocking terminology. NOT for implementation task breakdown. NOT for generic technology comparisons or best-practice research; use researching-web. NOT for docs updates; use documenting-code.
development33
looking-up-docs
Find current, version-correct library/API/framework docs through one lookup workflow. Use when the user says "look up docs", "how to use", "API for", "syntax for", "examples of", "show me the docs", mentions "ctx7"/"Context7", passes a `/org/project` library ID, or wants the latest/current/actual behavior of a library, framework, CLI, or API. NOT for comparisons, best-practice surveys, or recent ecosystem news — use researching-web.
tools33
operating-infra
Author, inspect, troubleshoot, and review infrastructure across IaC, Kubernetes, cloud resources, containers, CI/CD, and Linux hosts. Use when changing Terraform/OpenTofu, Kubernetes, Helm, Kustomize, Dockerfiles, GitHub Actions, AWS, GCP, Cloud Run, BigQuery, IAM, logs, instances, or service health. NOT for deploy/apply/rollback workflows (see deploying-infra). NOT for shell scripts or generic command pipelines (see writing-shell).
testing33
refactoring-code
Batch behavior-preserving refactors for multi-file, repeated-pattern, large-file, rename, move, extract, split, or restructure work. Use for "refactor across files", "batch rename", "update pattern everywhere", large files (500+ lines), or 5+ coordinated edits in one file. NOT for single targeted edits, behavior changes or bug fixes (use fixing-code), test-only refactors (use improving-tests), code review (use reviewing-code), or architecture redesign (use architecture-design/review).
development33
playwright-skill
Support-only Playwright runtime/reference for browser-automation — dev-server detection, a Node.js script runner, quiet screenshot helpers, SPA readiness helpers, and custom HTTP headers. Use when browser-automation selects the bundled Playwright fallback; do not route user intent here directly.
tools33
sequential-thinking
Structured stepwise reasoning with explicit revisions and branches. Use when the user says "think step by step", "sequential thinking", "plan this out", "reason through this", "branch this idea", or when tackling a hard multi-step problem (architecture decisions, ambiguous bugs, multi-constraint tradeoffs, plans that may need revision). NOT for trivial lookups, single-tool fetches, or tasks the model can answer directly without planning.
tools33
improving-tests
Improve test design and coverage with behavior-focused tests, useful seams, characterization tests, TDD, and test refactoring. Use when improving tests, adding coverage, refactoring brittle tests, removing test waste, or working test-first. NOT for fixing production bugs (use fixing-code), production-code refactors (use refactoring-code), or reviewing non-test code quality (use reviewing-code).
development33
cleanup-git
Remove merged local branches and stale git worktrees. Use when the user says "cleanup branches", "prune worktrees", "tidy git", "remove merged branches", "delete merged branches", "gone branches", or wants to clean local git state. NOT for creating commits, creating worktrees, or configuring git hooks.
development33
looking-up-docs
Find current, version-correct library/API/framework docs through one lookup workflow. Use when the user says "look up docs", "how to use", "API for", "syntax for", "examples of", "show me the docs", mentions "ctx7"/"Context7", passes a `/org/project` library ID, or wants the latest/current/actual behavior of a library, framework, CLI, or API. NOT for comparisons, best-practice surveys, or recent ecosystem news — use researching-web.
tools33
sequential-thinking
Structured stepwise reasoning with explicit revisions and branches. Use when the user says "think step by step", "sequential thinking", "plan this out", "reason through this", "branch this idea", or when tackling a hard multi-step problem (architecture decisions, ambiguous bugs, multi-constraint tradeoffs, plans that may need revision). NOT for trivial lookups, single-tool fetches, or tasks the model can answer directly without planning.
tools33
documenting-code
Create or update human-facing docs, agent-facing instructions, architecture docs, API docs, README content, and useful code comments from implementation facts. Use when docs are stale, missing, or must reflect code changes. NOT for code-quality review, prompt scoring, speculative docs, or ADRs unless explicitly requested.
development33
writing-shell
Idiomatic shell development for POSIX sh, Bash, Zsh, Fish, hooks, CI shell steps, and scriptable CLI glue. Use when writing or changing `.sh`, `.bash`, `.zsh`, `.fish`, `.bats`, shell functions, shell pipelines, or command-runner recipes. Emphasizes portability, quoting, safe filesystem/process handling, non-TUI CLI tools, ShellCheck, shfmt, Bats, and ShellSpec. NOT for Python, TypeScript, Go, web code, or infrastructure operations.
tools33
using-git-worktrees
Creates isolated git worktrees for parallel development. Use when starting feature work needing isolation or working on multiple branches simultaneously. NOT for simple branch switching, bulk branch cleanup (use cleanup-git), or git hook/config setup (use configuring-git-hygiene).
development33
spec-flow
Use when planning, executing, checkpointing, finishing, or inspecting lightweight spec-driven work. Runs one task at a time using `.spec/` markdown files and the bundled `specctl` helper. NOT for broad product discovery beyond a short requirement interview.
tools33
reviewing-instructions
Use when asked to lint, audit, review, or score AI-facing instruction files such as SKILL.md, AGENT.md, AGENTS.md, CLAUDE.md, platform body.md files, prompt files, rules, policies, and agent-facing references. NOT for application code review, harness configuration review, ordinary docs, tests, or generated build output.
development33
reviewing-code
Use when reviewing changed code, PRs, diffs, or specific files. Finds evidence-backed defects in security, correctness, tests, reliability, performance, maintainability, and docs. Supports quick, standard, deep, team, and external-review modes. NOT for repo-wide architecture review, general codebase exploration, fixing issues (use fixing-code), improving tests without a code review (use improving-tests), or applying refactors (use refactoring-code).
development33
refactoring-code
Batch behavior-preserving refactors for multi-file, repeated-pattern, large-file, rename, move, extract, split, or restructure work. Use for "refactor across files", "batch rename", "update pattern everywhere", large files (500+ lines), or 5+ coordinated edits in one file. NOT for single targeted edits, behavior changes or bug fixes (use fixing-code), test-only refactors (use improving-tests), code review (use reviewing-code), or architecture redesign (use architecture-design/review).
development33
operating-infra
Author, inspect, troubleshoot, and review infrastructure across IaC, Kubernetes, cloud resources, containers, CI/CD, and Linux hosts. Use when changing Terraform/OpenTofu, Kubernetes, Helm, Kustomize, Dockerfiles, GitHub Actions, AWS, GCP, Cloud Run, BigQuery, IAM, logs, instances, or service health. NOT for deploy/apply/rollback workflows (see deploying-infra). NOT for shell scripts or generic command pipelines (see writing-shell).
testing33
playwright-skill
Support-only Playwright runtime/reference for browser-automation — dev-server detection, a Node.js script runner, quiet screenshot helpers, SPA readiness helpers, and custom HTTP headers. Use when browser-automation selects the bundled Playwright fallback; do not route user intent here directly.
tools33
improving-tests
Improve test design and coverage with behavior-focused tests, useful seams, characterization tests, TDD, and test refactoring. Use when improving tests, adding coverage, refactoring brittle tests, removing test waste, or working test-first. NOT for fixing production bugs (use fixing-code), production-code refactors (use refactoring-code), or reviewing non-test code quality (use reviewing-code).
development33
looking-up-docs
Find current, version-correct library/API/framework docs through one lookup workflow. Use when the user says "look up docs", "how to use", "API for", "syntax for", "examples of", "show me the docs", mentions "ctx7"/"Context7", passes a `/org/project` library ID, or wants the latest/current/actual behavior of a library, framework, CLI, or API. NOT for comparisons, best-practice surveys, or recent ecosystem news — use researching-web.
tools33
fixing-code
Fix code defects with a reproducible feedback loop, root-cause diagnosis, minimal patch, regression test, and clean verification. Use when debugging, diagnosing, or resolving lint/test/build failures. NOT for behavior-preserving refactors (use refactoring-code), test-suite cleanup without a production bug (use improving-tests), or code review findings without fixes (use reviewing-code).
development33
documenting-code
Create or update human-facing docs, agent-facing instructions, architecture docs, API docs, README content, and useful code comments from implementation facts. Use when docs are stale, missing, or must reflect code changes. NOT for code-quality review, prompt scoring, speculative docs, or ADRs unless explicitly requested.
development33
evolving-config
Audit and improve AI coding-agent configuration. Use when reviewing or changing Claude Code, Pi, Codex, skill, agent, hook, MCP, permission, package, or generated-export setup. Default is review-only; fixes require explicit user approval or --fix. NOT for application config, git hygiene, code bugs, ordinary docs, or generated files without their source.
tools33
configuring-git-hygiene
Configure safe git workflow hygiene: pre-commit/pre-push hooks, Gitleaks secret scanning, .gitignore rules, local git config, and guardrails. Use when setting up git hooks, gitleaks/git leaks, staged pre-commit checks, pre-push validation, core.hooksPath, .gitignore, or git config best practices. NOT for creating commits (use committing-code), cleaning branches/worktrees (use cleanup-git), or creating worktrees (use using-git-worktrees).
development33
committing-code
Create normal git commits with logical grouping. Use when committing, saving changes, creating commits, or grouping work into commits. NOT for amending, rebasing, force-pushing, or rewriting history.
development33
deploying-infra
Validate infrastructure changes and, after explicit confirmation, apply Terraform, Helm, Kustomize, or Kubernetes deployments. Use when the user says "deploy", "deploy to staging", "terraform apply", "helm upgrade", "kubectl apply", "rollout", "deploy check", "validate deployment", or "validate infrastructure". Dockerfiles and GitHub Actions are validate-only here. NOT for ongoing service troubleshooting, cloud inspection, rollback investigation, or authoring infra from scratch; use operating-infra for those.
development33
browser-automation
Browser automation for rendered UI exploration, validation, screenshots, recordings, and end-to-end flows. Use when a task needs an actual browser or rendered DOM: inspect UI state, click/fill forms, debug frontend behavior, capture evidence, verify a feature, or run/generate browser tests. NOT for API checks or pure logic tests where curl, unit tests, or JSDOM is cheaper.
tools33
brainstorming-ideas
Brainstorm ideas and stress-test draft plans before coding. Use when brainstorming, exploring approaches, designing a feature/API/flow, grilling or debating a bounded plan, challenging assumptions, or resolving design-blocking terminology. NOT for implementation task breakdown. NOT for generic technology comparisons or best-practice research; use researching-web. NOT for docs updates; use documenting-code.
development33
writing-web
Simple web development with HTML, CSS, JS, and HTMX. Use when working with .html, .css, or .htmx files, web templates, stylesheets, or vanilla JS scripts. NOT for React/Vue/Angular (use writing-typescript) or Node.js backends.
development33
writing-shell
Idiomatic shell development for POSIX sh, Bash, Zsh, Fish, hooks, CI shell steps, and scriptable CLI glue. Use when writing or changing `.sh`, `.bash`, `.zsh`, `.fish`, `.bats`, shell functions, shell pipelines, or command-runner recipes. Emphasizes portability, quoting, safe filesystem/process handling, non-TUI CLI tools, ShellCheck, shfmt, Bats, and ShellSpec. NOT for Python, TypeScript, Go, web code, or infrastructure operations.
tools33
using-git-worktrees
Creates isolated git worktrees for parallel development. Use when starting feature work needing isolation or working on multiple branches simultaneously. NOT for simple branch switching, bulk branch cleanup (use cleanup-git), or git hook/config setup (use configuring-git-hygiene).
development33
spec-flow
Use when planning, executing, checkpointing, finishing, or inspecting lightweight spec-driven work. Runs one task at a time using `.spec/` markdown files and the bundled `specctl` helper. NOT for broad product discovery beyond a short requirement interview.
tools33
reviewing-instructions
Use when asked to lint, audit, review, or score AI-facing instruction files such as SKILL.md, AGENT.md, AGENTS.md, CLAUDE.md, platform body.md files, prompt files, rules, policies, and agent-facing references. NOT for application code review, harness configuration review, ordinary docs, tests, or generated build output.
development33
reviewing-code
Use when reviewing changed code, PRs, diffs, or specific files. Finds evidence-backed defects in security, correctness, tests, reliability, performance, maintainability, and docs. Supports quick, standard, deep, team, and external-review modes. NOT for repo-wide architecture review, general codebase exploration, fixing issues (use fixing-code), improving tests without a code review (use improving-tests), or applying refactors (use refactoring-code).
development33
researching-web
Web research via platform web tools. Use for technical comparisons, recent facts, ecosystem news, best practices, standards, or questions needing grounded web evidence. NOT for API syntax lookup or code examples — use looking-up-docs for those. NOT for repo-specific questions — search local files first.
tools33
playwright-skill
Support-only Playwright runtime/reference for browser-automation — dev-server detection, a Node.js script runner, quiet screenshot helpers, SPA readiness helpers, and custom HTTP headers. Use when browser-automation selects the bundled Playwright fallback; do not route user intent here directly.
tools33
refactoring-code
Batch behavior-preserving refactors for multi-file, repeated-pattern, large-file, rename, move, extract, split, or restructure work. Use for "refactor across files", "batch rename", "update pattern everywhere", large files (500+ lines), or 5+ coordinated edits in one file. NOT for single targeted edits, behavior changes or bug fixes (use fixing-code), test-only refactors (use improving-tests), code review (use reviewing-code), or architecture redesign (use architecture-design/review).
development33
operating-infra
Author, inspect, troubleshoot, and review infrastructure across IaC, Kubernetes, cloud resources, containers, CI/CD, and Linux hosts. Use when changing Terraform/OpenTofu, Kubernetes, Helm, Kustomize, Dockerfiles, GitHub Actions, AWS, GCP, Cloud Run, BigQuery, IAM, logs, instances, or service health. NOT for deploy/apply/rollback workflows (see deploying-infra). NOT for shell scripts or generic command pipelines (see writing-shell).
testing33
looking-up-docs
Find current, version-correct library/API/framework docs through one lookup workflow. Use when the user says "look up docs", "how to use", "API for", "syntax for", "examples of", "show me the docs", mentions "ctx7"/"Context7", passes a `/org/project` library ID, or wants the latest/current/actual behavior of a library, framework, CLI, or API. NOT for comparisons, best-practice surveys, or recent ecosystem news — use researching-web.
tools33
improving-tests
Improve test design and coverage with behavior-focused tests, useful seams, characterization tests, TDD, and test refactoring. Use when improving tests, adding coverage, refactoring brittle tests, removing test waste, or working test-first. NOT for fixing production bugs (use fixing-code), production-code refactors (use refactoring-code), or reviewing non-test code quality (use reviewing-code).
development33
evolving-config
Audit and improve AI coding-agent configuration. Use when reviewing or changing Claude Code, Pi, Codex, skill, agent, hook, MCP, permission, package, or generated-export setup. Default is review-only; fixes require explicit user approval or --fix. NOT for application config, git hygiene, code bugs, ordinary docs, or generated files without their source.
tools33
fixing-code
Fix code defects with a reproducible feedback loop, root-cause diagnosis, minimal patch, regression test, and clean verification. Use when debugging, diagnosing, or resolving lint/test/build failures. NOT for behavior-preserving refactors (use refactoring-code), test-suite cleanup without a production bug (use improving-tests), or code review findings without fixes (use reviewing-code).
development33
configuring-git-hygiene
Configure safe git workflow hygiene: pre-commit/pre-push hooks, Gitleaks secret scanning, .gitignore rules, local git config, and guardrails. Use when setting up git hooks, gitleaks/git leaks, staged pre-commit checks, pre-push validation, core.hooksPath, .gitignore, or git config best practices. NOT for creating commits (use committing-code), cleaning branches/worktrees (use cleanup-git), or creating worktrees (use using-git-worktrees).
development33
committing-code
Create normal git commits with logical grouping. Use when committing, saving changes, creating commits, or grouping work into commits. NOT for amending, rebasing, force-pushing, or rewriting history.
development33
cleanup-git
Remove merged local branches and stale git worktrees. Use when the user says "cleanup branches", "prune worktrees", "tidy git", "remove merged branches", "delete merged branches", "gone branches", or wants to clean local git state. NOT for creating commits, creating worktrees, or configuring git hooks.
development33
browser-automation
Browser automation for rendered UI exploration, validation, screenshots, recordings, and end-to-end flows. Use when a task needs an actual browser or rendered DOM: inspect UI state, click/fill forms, debug frontend behavior, capture evidence, verify a feature, or run/generate browser tests. NOT for API checks or pure logic tests where curl, unit tests, or JSDOM is cheaper.
tools33
writing-web
Simple web development with HTML, CSS, JS, and HTMX. Use when working with .html, .css, or .htmx files, web templates, stylesheets, or vanilla JS scripts. NOT for React/Vue/Angular (use writing-typescript) or Node.js backends.
development33
using-git-worktrees
Creates isolated git worktrees for parallel development. Use when starting feature work needing isolation or working on multiple branches simultaneously. NOT for simple branch switching, bulk branch cleanup (use cleanup-git), or git hook/config setup (use configuring-git-hygiene).
development33
writing-shell
Idiomatic shell development for POSIX sh, Bash, Zsh, Fish, hooks, CI shell steps, and scriptable CLI glue. Use when writing or changing `.sh`, `.bash`, `.zsh`, `.fish`, `.bats`, shell functions, shell pipelines, or command-runner recipes. Emphasizes portability, quoting, safe filesystem/process handling, non-TUI CLI tools, ShellCheck, shfmt, Bats, and ShellSpec. NOT for Python, TypeScript, Go, web code, or infrastructure operations.
tools33
reviewing-instructions
Use when asked to lint, audit, review, or score AI-facing instruction files such as SKILL.md, AGENT.md, AGENTS.md, CLAUDE.md, platform body.md files, prompt files, rules, policies, and agent-facing references. NOT for application code review, harness configuration review, ordinary docs, tests, or generated build output.
development33
spec-flow
Use when planning, executing, checkpointing, finishing, or inspecting lightweight spec-driven work. Runs one task at a time using `.spec/` markdown files and the bundled `specctl` helper. NOT for broad product discovery beyond a short requirement interview.
tools33
reviewing-code
Use when reviewing changed code, PRs, diffs, or specific files. Finds evidence-backed defects in security, correctness, tests, reliability, performance, maintainability, and docs. Supports quick, standard, deep, team, and external-review modes. NOT for repo-wide architecture review, general codebase exploration, fixing issues (use fixing-code), improving tests without a code review (use improving-tests), or applying refactors (use refactoring-code).
development33
writing-python
Idiomatic Python 3.12+ development. Use when writing Python code, CLI tools, scripts, or services. Emphasizes stdlib, type hints, uv/ruff/pyright toolchain, and minimal dependencies. NOT for Go, TypeScript, or shell-only tasks.
tools32
using-modern-cli
Prefer modern CLI tools for shell and file workflows — rg, fd, bat, eza, sd, dust, procs, and delta over legacy grep/find/cat/ls/sed/du/ps/diff. Use when writing bash scripts, optimizing command chains, or replacing legacy Unix tools. NOT for repo-wide code search, architecture review, AST/codegraph/GitNexus evidence, or application logic.
tools32
writing-go
Idiomatic Go development. Use when writing Go code, designing APIs, reviewing Go implementations, or changing Go tests. Follow the module's target Go version. Prefer stdlib, concrete types, explicit errors, context propagation, and behavior tests. NOT for Python, TypeScript, shell scripts, or infra-only work.
development32
exploring-repos
Explore public GitHub repositories in Pi using GitHub CLI, local clones, and web tools. Use when the user asks how a public repo works, wants architecture orientation, or needs repo-level Q&A. NOT for library API docs (use looking-up-docs) or local private codebases (use a local codebase workflow).
tools32
writing-python
Idiomatic Python 3.12+ development. Use when writing Python code, CLI tools, scripts, or services. Emphasizes stdlib, type hints, uv/ruff/pyright toolchain, and minimal dependencies. NOT for Go, TypeScript, or shell-only tasks.
tools32
exploring-repos
Explore public GitHub repositories in Pi using GitHub CLI, local clones, and web tools. Use when the user asks how a public repo works, wants architecture orientation, or needs repo-level Q&A. NOT for library API docs (use looking-up-docs) or local private codebases (use a local codebase workflow).
tools32
using-modern-cli
Prefer modern CLI tools for shell and file workflows — rg, fd, bat, eza, sd, dust, procs, and delta over legacy grep/find/cat/ls/sed/du/ps/diff. Use when writing bash scripts, optimizing command chains, or replacing legacy Unix tools. NOT for repo-wide code search, architecture review, AST/codegraph/GitNexus evidence, or application logic.
tools32
writing-go
Idiomatic Go development. Use when writing Go code, designing APIs, reviewing Go implementations, or changing Go tests. Follow the module's target Go version. Prefer stdlib, concrete types, explicit errors, context propagation, and behavior tests. NOT for Python, TypeScript, shell scripts, or infra-only work.
development32
committing-code
Create git commits with logical grouping. Use when committing, saving changes, creating commits, or grouping work into commits.
development32
smart-explore
Token-efficient local code navigation and extraction. Use when exploring a known file or bounded module outline, finding a known symbol in a scoped area, or extracting exact function/type bodies with smart_outline, smart_search, and smart_unfold. NOT for repo-wide structural pattern search, architecture or trace-flow questions, ast-grep/codegraph/GitNexus evidence, or broad caller/implementation maps.
development32
writing-typescript
Idiomatic TypeScript development. Use when writing TypeScript code, Node.js services, React apps, or TypeScript design advice. Emphasizes strict typing, boundary validation, composition, behavior tests, and project-configured tooling. NOT for Go, Python, plain HTML/CSS/JS, or server-rendered templates.
tools32
writing-go
Idiomatic Go development. Use when writing Go code, designing APIs, reviewing Go implementations, or changing Go tests. Follow the module's target Go version. Prefer stdlib, concrete types, explicit errors, context propagation, and behavior tests. NOT for Python, TypeScript, shell scripts, or infra-only work.
development32
writing-python
Idiomatic Python 3.12+ development. Use when writing Python code, CLI tools, scripts, or services. Emphasizes stdlib, type hints, uv/ruff/pyright toolchain, and minimal dependencies. NOT for Go, TypeScript, or shell-only tasks.
tools32
smart-explore
Token-efficient local code navigation and extraction. Use when exploring a known file or bounded module outline, finding a known symbol in a scoped area, or extracting exact function/type bodies with smart_outline, smart_search, and smart_unfold. NOT for repo-wide structural pattern search, architecture or trace-flow questions, ast-grep/codegraph/GitNexus evidence, or broad caller/implementation maps.
development32
committing-code
Create git commits with logical grouping. Use when committing, saving changes, creating commits, or grouping work into commits.
development32
playwright-skill
Playwright primitives for real-browser automation — dev-server detection, a Node.js script runner, and helpers for clicks, form fills, screenshots, multi-viewport, custom HTTP headers. Use when a task needs an actual browser (rendered DOM, visual checks, multi-page flows, cross-browser behavior). Not for API tests or logic tests where curl or JSDOM is cheaper.
tools32
playwright-skill
Playwright primitives for real-browser automation — dev-server detection, a Node.js script runner, and helpers for clicks, form fills, screenshots, multi-viewport, custom HTTP headers. Use when a task needs an actual browser (rendered DOM, visual checks, multi-page flows, cross-browser behavior). Not for API tests or logic tests where curl or JSDOM is cheaper.
tools32
writing-typescript
Idiomatic TypeScript development. Use when writing TypeScript code, Node.js services, React apps, or TypeScript design advice. Emphasizes strict typing, boundary validation, composition, behavior tests, and project-configured tooling. NOT for Go, Python, plain HTML/CSS/JS, or server-rendered templates.
tools32
committing-code
Create git commits with logical grouping. Use when committing, saving changes, creating commits, or grouping work into commits.
development32
browser-automation
Browser automation for rendered UI exploration, validation, screenshots, recordings, and end-to-end flows. Use when a task needs an actual browser or rendered DOM: inspect UI state, click/fill forms, debug frontend behavior, capture evidence, verify a feature, or run/generate browser tests. NOT for API checks or pure logic tests where curl, unit tests, or JSDOM is cheaper.
tools32
reviewing-code
Code review covering security, correctness, quality, tests, implementation, and documentation. Use when the user asks to review code, check changes, audit a PR or diff, or find line-level refactoring opportunities. NOT for repo-wide architecture review, codebase analysis, fixing issues (use fixing-code), or applying refactors (use refactoring-code).
development32
reviewing-code
Code review covering security, correctness, quality, tests, implementation, and documentation. Use when the user asks to review code, check changes, audit a PR or diff, or find line-level refactoring opportunities. NOT for repo-wide architecture review, codebase analysis, fixing issues (use fixing-code), or applying refactors (use refactoring-code).
development32
playwright-skill
Playwright primitives for real-browser automation — dev-server detection, a Node.js script runner, and helpers for clicks, form fills, screenshots, multi-viewport, custom HTTP headers. Use when a task needs an actual browser (rendered DOM, visual checks, multi-page flows, cross-browser behavior). Not for API tests or logic tests where curl or JSDOM is cheaper.
tools32
writing-typescript
Idiomatic TypeScript development. Use when writing TypeScript code, Node.js services, React apps, or TypeScript design advice. Emphasizes strict typing, boundary validation, composition, behavior tests, and project-configured tooling. NOT for Go, Python, plain HTML/CSS/JS, or server-rendered templates.
tools32
writing-python
Idiomatic Python 3.12+ development. Use when writing Python code, CLI tools, scripts, or services. Emphasizes stdlib, type hints, uv/ruff/pyright toolchain, and minimal dependencies. NOT for Go, TypeScript, or shell-only tasks.
tools32
reviewing-code
Code review covering security, correctness, quality, tests, implementation, and documentation. Use when the user asks to review code, check changes, audit a PR or diff, or find line-level refactoring opportunities. NOT for repo-wide architecture review, codebase analysis, fixing issues (use fixing-code), or applying refactors (use refactoring-code).
development32
playwright-skill
Playwright primitives for real-browser automation — dev-server detection, a Node.js script runner, and helpers for clicks, form fills, screenshots, multi-viewport, custom HTTP headers. Use when a task needs an actual browser (rendered DOM, visual checks, multi-page flows, cross-browser behavior). Not for API tests or logic tests where curl or JSDOM is cheaper.
tools32
reviewing-code
Code review covering security, correctness, quality, tests, implementation, and documentation. Use when the user asks to review code, check changes, audit a PR or diff, or find line-level refactoring opportunities. NOT for repo-wide architecture review, codebase analysis, fixing issues (use fixing-code), or applying refactors (use refactoring-code).
development32
exploring-repos
Explore public GitHub repositories in Pi using GitHub CLI, local clones, and web tools. Use when the user asks how a public repo works, wants architecture orientation, or needs repo-level Q&A. NOT for library API docs (use looking-up-docs) or local private codebases (use a local codebase workflow).
tools32
smart-explore
Token-efficient local code navigation and extraction. Use when exploring a known file or bounded module outline, finding a known symbol in a scoped area, or extracting exact function/type bodies with smart_outline, smart_search, and smart_unfold. NOT for repo-wide structural pattern search, architecture or trace-flow questions, ast-grep/codegraph/GitNexus evidence, or broad caller/implementation maps.
development32
writing-typescript
Idiomatic TypeScript development. Use when writing TypeScript code, Node.js services, React apps, or TypeScript design advice. Emphasizes strict typing, boundary validation, composition, behavior tests, and project-configured tooling. NOT for Go, Python, plain HTML/CSS/JS, or server-rendered templates.
tools32
writing-go
Idiomatic Go development. Use when writing Go code, designing APIs, reviewing Go implementations, or changing Go tests. Follow the module's target Go version. Prefer stdlib, concrete types, explicit errors, context propagation, and behavior tests. NOT for Python, TypeScript, shell scripts, or infra-only work.
development32
using-modern-cli
Prefer modern CLI tools for shell and file workflows — rg, fd, bat, eza, sd, dust, procs, and delta over legacy grep/find/cat/ls/sed/du/ps/diff. Use when writing bash scripts, optimizing command chains, or replacing legacy Unix tools. NOT for repo-wide code search, architecture review, AST/codegraph/GitNexus evidence, or application logic.
tools32
smart-explore
Token-efficient local code navigation and extraction. Use when exploring a known file or bounded module outline, finding a known symbol in a scoped area, or extracting exact function/type bodies with smart_outline, smart_search, and smart_unfold. NOT for repo-wide structural pattern search, architecture or trace-flow questions, ast-grep/codegraph/GitNexus evidence, or broad caller/implementation maps.
development32
writing-python
Idiomatic Python 3.12+ development. Use when writing Python code, CLI tools, scripts, or services. Emphasizes stdlib, type hints, uv/ruff/pyright toolchain, and minimal dependencies. NOT for Go, TypeScript, or shell-only tasks.
tools32
using-modern-cli
Prefer modern CLI tools for shell and file workflows — rg, fd, bat, eza, sd, dust, procs, and delta over legacy grep/find/cat/ls/sed/du/ps/diff. Use when writing bash scripts, optimizing command chains, or replacing legacy Unix tools. NOT for repo-wide code search, architecture review, AST/codegraph/GitNexus evidence, or application logic.
tools32
writing-go
Idiomatic Go development. Use when writing Go code, designing APIs, reviewing Go implementations, or changing Go tests. Follow the module's target Go version. Prefer stdlib, concrete types, explicit errors, context propagation, and behavior tests. NOT for Python, TypeScript, shell scripts, or infra-only work.
development32
committing-code
Create git commits with logical grouping. Use when committing, saving changes, creating commits, or grouping work into commits.
development32
writing-typescript
Idiomatic TypeScript development. Use when writing TypeScript code, Node.js services, React apps, or TypeScript design advice. Emphasizes strict typing, boundary validation, composition, behavior tests, and project-configured tooling. NOT for Go, Python, plain HTML/CSS/JS, or server-rendered templates.
tools32
smart-explore
Token-efficient local code navigation and extraction. Use when exploring a known file or bounded module outline, finding a known symbol in a scoped area, or extracting exact function/type bodies with smart_outline, smart_search, and smart_unfold. NOT for repo-wide structural pattern search, architecture or trace-flow questions, ast-grep/codegraph/GitNexus evidence, or broad caller/implementation maps.
development32
exploring-repos
Explore public GitHub repositories in Pi using GitHub CLI, local clones, and web tools. Use when the user asks how a public repo works, wants architecture orientation, or needs repo-level Q&A. NOT for library API docs (use looking-up-docs) or local private codebases (use a local codebase workflow).
tools32
reviewing-instructions
Review and score AI agent/skill instruction files for quality — signal density, scope specificity, output structure, failure handling, and routing precision. Use when asked to "lint", "audit", "review", or "score" prompts, SKILL.md, AGENT.md, AGENTS.md, CLAUDE.md, platform-specific body.md, reference markdown, or other markdown files explicitly meant to be read by AI agents.
development32
exploring-repos
Explore public GitHub repositories in Pi using GitHub CLI, local clones, and web tools. Use when the user asks how a public repo works, wants architecture orientation, or needs repo-level Q&A. NOT for library API docs (use looking-up-docs) or local private codebases (use a local codebase workflow).
tools32
cleanup-git
Remove merged local branches and stale git worktrees. Use when the user says "cleanup branches", "prune worktrees", "tidy git", "remove merged branches", "gone branches", or wants to clean local git state.
development32
cleanup-git
Remove merged local branches and stale git worktrees. Use when the user says "cleanup branches", "prune worktrees", "tidy git", "remove merged branches", "gone branches", or wants to clean local git state.
development32
reviewing-instructions
Review and score AI agent/skill instruction files for quality — signal density, scope specificity, output structure, failure handling, and routing precision. Use when asked to "lint", "audit", "review", or "score" prompts, SKILL.md, AGENT.md, AGENTS.md, CLAUDE.md, platform-specific body.md, reference markdown, or other markdown files explicitly meant to be read by AI agents.
development32
using-modern-cli
Prefer modern CLI tools for shell and file workflows — rg, fd, bat, eza, sd, dust, procs, and delta over legacy grep/find/cat/ls/sed/du/ps/diff. Use when writing bash scripts, optimizing command chains, or replacing legacy Unix tools. NOT for repo-wide code search, architecture review, AST/codegraph/GitNexus evidence, or application logic.
tools32
using-cloud-cli
Cloud CLI patterns for GCP and AWS. Use when running bq queries, gcloud commands, aws commands, or making decisions about cloud services. Covers BigQuery cost optimization and operational best practices. NOT for Terraform or Kubernetes architectural decisions (see managing-infra).
tools31
spec-interview
Capture PRD-quality requirements through structured Q&A. Use when a new requirement needs deep exploration — produces a `REQ-*.md` via 8–15 targeted questions. NOT for creating tasks or implementation plans — use spec-plan for that.
testing31
parsing-documents
Extract structured data from PDF documents — text, tables, forms, and metadata. Use when reading or extracting content from a `.pdf` file, parsing invoices/reports/scanned documents, or converting PDF data to JSON/CSV. NOT for generating PDFs, and NOT for plain-text/markdown files (read those directly).
documentation31
looking-up-docs
Find current, factual library/API/framework documentation through a tool-fallback chain. Use when the user says "look up docs", "how to use", "API for", "syntax for", "examples of", "show me the docs", or wants the latest/current/actual behavior of a library, framework, CLI, or API. NOT for comparisons, best-practice surveys, or recent ecosystem news — use researching-web. NOT for raw ctx7 CLI mechanics — that is context7-cli.
tools31
reviewing-cc-config
Review Claude Code configuration for context efficiency, signal density, and anti-patterns. Use when user says "review config", "review setup", "check configuration", "review cc config", "context review", "config review", "review my setup", "review skills", "review agents", "review hooks", or wants feedback on their Claude Code configuration quality. NOT for editing config files — review only; user applies fixes unless --fix is passed. NOT for applying config changes (use `evolving-config`).
development31
writing-python
Idiomatic Python 3.12+ development. Use when writing Python code, CLI tools, scripts, or services. Emphasizes stdlib, type hints, uv/ruff/pyright toolchain, and minimal dependencies. NOT for Go, TypeScript, or shell-only tasks.
tools31
writing-web
Simple web development with HTML, CSS, JS, and HTMX. Use when working with .html, .css, or .htmx files, web templates, stylesheets, or vanilla JS scripts. NOT for React/Vue/Angular (use writing-typescript) or Node.js backends.
development31
brainstorming-ideas
Brainstorm ideas and stress-test draft plans before coding. Use when brainstorming, exploring approaches, designing a feature/API/flow, grilling a plan, challenging assumptions, or resolving terminology that blocks the design. NOT for implementation task breakdown; use the spec-plan skill. NOT for general documentation updates; use documenting-code or learning-patterns.
development31
context7-cli
ctx7 (Context7) CLI mechanics for querying versioned library documentation. Use when the user mentions "ctx7" or "context7", passes a `/org/project` library ID, or another skill needs the exact ctx7 command workflow. NOT the docs-lookup decision flow or web fallback — that is looking-up-docs.
tools31
evolving-config
Audit AI coding-agent configuration against current features and local usage. Use when the user wants to improve Claude Code, Pi, Codex, Gemini, skill, hook, or agent configuration. NOT for writing new application code, fixing bugs, or any task that isn't about agent/tool configuration files. NOT for review-only audits without applying changes (use `reviewing-cc-config`).
tools31
mem-history
Query project history, past decisions, and known gotchas from claude-mem observations. Use when user asks "last session", "did we already", "what did we decide", "project history", "timeline", or "what happened with".
development31
researching-web
Web research via Perplexity and platform web tools. Use for technical comparisons, recent facts, ecosystem news, best practices, standards, or questions needing grounded web evidence. NOT for API syntax lookup or code examples — use looking-up-docs for those. NOT for repo-specific questions — search local files first.
tools31
documenting-code
Update project documentation based on code changes. Use when the user asks to update docs, document behavior, add README content, or align docs with recent implementation changes. NOT for extracting session learnings or authoring ADRs (use learning-patterns) or code-quality feedback (use reviewing-code).
development31
using-git-worktrees
Creates isolated git worktrees for parallel development. Use when starting feature work needing isolation or working on multiple branches simultaneously. Not for simple branch switching or basic git operations.
development31
improving-codebase-architecture
Find deepening opportunities in a codebase, informed by the domain language in CONTEXT.md and the decisions in docs/adr/. Use when the user wants to improve architecture, find refactoring opportunities, consolidate tightly-coupled modules, or make a codebase more testable and AI-navigable. NOT for line-level cleanup (use reviewing-code) or batch edits (use refactoring-code).
development31
refactoring-code
Batch refactoring via MorphLLM edit_file. Use for "refactor across files", "batch rename", "update pattern everywhere", large files (500+ lines), 5+ edits in same file, or applying an approved architecture-deepening refactor. NOT for single-file targeted edits (use built-in Edit) or code review (use reviewing-code).
development31
writing-python
Idiomatic Python 3.12+ development. Use when writing Python code, CLI tools, scripts, or services. Emphasizes stdlib, type hints, uv/ruff/pyright toolchain, and minimal dependencies. NOT for Go, TypeScript, or shell-only tasks.
tools31
reviewing-code
Code review covering security, quality, tests, implementation, documentation, and architecture / module-depth. Use when the user asks to review code, check changes, audit a PR or diff, find refactoring opportunities, or look for shallow modules and over-abstraction. NOT for fixing the issues found (use fixing-code) or applying refactors (use refactoring-code).
development31
using-cloud-cli
Cloud CLI patterns for GCP and AWS. Use when running bq queries, gcloud commands, aws commands, or making decisions about cloud services. Covers BigQuery cost optimization and operational best practices. NOT for Terraform or Kubernetes architectural decisions (see managing-infra).
tools31
managing-infra
Infrastructure patterns for Kubernetes, Terraform, Helm, Kustomize, and GitHub Actions. Use when making K8s architectural decisions, choosing between Helm vs Kustomize, structuring Terraform modules, writing CI/CD workflows, or applying security best practices. NOT for cloud CLI commands (see using-cloud-cli) or deploy validation and apply workflows (see deploying-infra).
tools31
spec-plan
Turn a requirement or a concrete feature idea into an EPIC with vertical-slice TASKs. Use when you have a REQ file, or a feature idea already specific enough for a quick 3–5 question pass, and need an executable plan with dependencies and acceptance criteria. NOT for open-ended idea exploration — use brainstorming-ideas. NOT for capturing PRD-quality requirements — use spec-interview. NOT for implementing tasks — use spec-work.
testing31
spec-work
Implement the next ready task. Use when starting a development session — selects the highest-priority ready task, plans with a specialist subagent, implements with approval at every step, verifies quality gates, and commits. One task per session. NOT for batch task execution or planning new work — use spec-plan for planning.
development31
spec-status
Spec-driven development status and orientation. Use when checking overall project state, viewing a specific task with its linked req/epic, listing tasks by status, running a quality audit for orphans/cycles/missing fields, or for a pipeline overview when unsure which spec sub-skill to use. NOT for mutating state — read-only; use spec-done or spec-work for state changes.
development31
spec-new
Create a single TASK or REQ file from a template. Use for one-off artifact creation without the full planning workflow. NOT for full project bootstrap (spec-init) or multi-task planning from a requirement (spec-plan).
tools31
mem-history
Query project history, past decisions, and known gotchas from claude-mem observations. Use when user asks "last session", "did we already", "what did we decide", "project history", "timeline", or "what happened with".
development31
improving-tests
Improve test design and coverage, including TDD/red-green-refactor guidance. Use when improving tests, refactoring tests, adding coverage, using TDD, or removing test waste. NOT for fixing production bugs (use fixing-code) or reviewing non-test code quality (use reviewing-code).
development31
improving-codebase-architecture
Find deepening opportunities in a codebase, informed by the domain language in CONTEXT.md and the decisions in docs/adr/. Use when the user wants to improve architecture, find refactoring opportunities, consolidate tightly-coupled modules, or make a codebase more testable and AI-navigable. NOT for line-level cleanup (use reviewing-code) or batch edits (use refactoring-code).
development31
writing-web
Simple web development with HTML, CSS, JS, and HTMX. Use when working with .html, .css, or .htmx files, web templates, stylesheets, or vanilla JS scripts. NOT for React/Vue/Angular (use writing-typescript) or Node.js backends.
development31
watch-team
Monitor a Claude Code team in tmux, auto-approve prompts, and report status. Use when the user says "watch the team", "monitor agents", "auto-approve", "babysit team", or "agents are stuck". NOT for single-agent monitoring, non-tmux setups, or general process supervision.
development31
evolving-config
Audit AI coding-agent configuration against current features and local usage. Use when the user wants to improve Claude Code, Pi, Codex, Gemini, skill, hook, or agent configuration. NOT for writing new application code, fixing bugs, or any task that isn't about agent/tool configuration files. NOT for review-only audits without applying changes (use `reviewing-cc-config`).
tools31
testing-e2e
Playwright-based end-to-end browser testing of user flows. Use when running existing Playwright tests, generating browser checks, recording a visible session, or verifying a user journey end-to-end. NOT for unit tests, API-only tests, or logic tests where curl or JSDOM suffices — use improving-tests or fixing-code instead.
development31
spec-interview
Capture PRD-quality requirements through structured Q&A. Use when a new requirement needs deep exploration — produces a `REQ-*.md` via 8–15 targeted questions. NOT for creating tasks or implementation plans — use spec-plan for that.
testing31
spec-interview
Capture PRD-quality requirements through structured Q&A. Use when a new requirement needs deep exploration — produces a `REQ-*.md` via 8–15 targeted questions. NOT for creating tasks or implementation plans — use spec-plan for that.
testing31
using-cloud-cli
Cloud CLI patterns for GCP and AWS. Use when running bq queries, gcloud commands, aws commands, or making decisions about cloud services. Covers BigQuery cost optimization and operational best practices. NOT for Terraform or Kubernetes architectural decisions (see managing-infra).
tools31
refactoring-code
Batch refactoring via MorphLLM edit_file. Use for "refactor across files", "batch rename", "update pattern everywhere", large files (500+ lines), 5+ edits in same file, or applying an approved architecture-deepening refactor. NOT for single-file targeted edits (use built-in Edit) or code review (use reviewing-code).
development31
researching-web
Web research via Perplexity and platform web tools. Use for technical comparisons, recent facts, ecosystem news, best practices, standards, or questions needing grounded web evidence. NOT for API syntax lookup or code examples — use looking-up-docs for those. NOT for repo-specific questions — search local files first.
tools31
context7-cli
ctx7 (Context7) CLI mechanics for querying versioned library documentation. Use when the user mentions "ctx7" or "context7", passes a `/org/project` library ID, or another skill needs the exact ctx7 command workflow. NOT the docs-lookup decision flow or web fallback — that is looking-up-docs.
tools31
watch-team
Monitor a Claude Code team in tmux, auto-approve prompts, and report status. Use when the user says "watch the team", "monitor agents", "auto-approve", "babysit team", or "agents are stuck". NOT for single-agent monitoring, non-tmux setups, or general process supervision.
development31
writing-typescript
Idiomatic TypeScript development. Use when writing TypeScript code, Node.js services, React apps, or discussing TS patterns. Emphasizes strict typing, composition, and modern tooling (bun/vite). NOT for Go, Python, plain HTML/CSS/JS, or server-rendered templates (use writing-web for those).
tools31
parsing-documents
Extract structured data from PDF documents — text, tables, forms, and metadata. Use when reading or extracting content from a `.pdf` file, parsing invoices/reports/scanned documents, or converting PDF data to JSON/CSV. NOT for generating PDFs, and NOT for plain-text/markdown files (read those directly).
documentation31
managing-infra
Infrastructure patterns for Kubernetes, Terraform, Helm, Kustomize, and GitHub Actions. Use when making K8s architectural decisions, choosing between Helm vs Kustomize, structuring Terraform modules, writing CI/CD workflows, or applying security best practices. NOT for cloud CLI commands (see using-cloud-cli) or deploy validation and apply workflows (see deploying-infra).
tools31
context7-cli
ctx7 (Context7) CLI mechanics for querying versioned library documentation. Use when the user mentions "ctx7" or "context7", passes a `/org/project` library ID, or another skill needs the exact ctx7 command workflow. NOT the docs-lookup decision flow or web fallback — that is looking-up-docs.
tools31
fixing-code
Fix code problems with disciplined diagnosis — run checks, build a repro for bugs, rank falsifiable hypotheses, fix one issue at a time, and verify until clean. Use when fixing, debugging, diagnosing, or resolving lint/test/build failures.
development31
documenting-code
Update project documentation based on code changes. Use when the user asks to update docs, document behavior, add README content, or align docs with recent implementation changes. NOT for extracting session learnings or authoring ADRs (use learning-patterns) or code-quality feedback (use reviewing-code).
development31
managing-infra
Infrastructure patterns for Kubernetes, Terraform, Helm, Kustomize, and GitHub Actions. Use when making K8s architectural decisions, choosing between Helm vs Kustomize, structuring Terraform modules, writing CI/CD workflows, or applying security best practices. NOT for cloud CLI commands (see using-cloud-cli) or deploy validation and apply workflows (see deploying-infra).
tools31
mem-history
Query project history, past decisions, and known gotchas from claude-mem observations. Use when user asks "last session", "did we already", "what did we decide", "project history", "timeline", or "what happened with".
development31
parsing-documents
Extract structured data from PDF documents — text, tables, forms, and metadata. Use when reading or extracting content from a `.pdf` file, parsing invoices/reports/scanned documents, or converting PDF data to JSON/CSV. NOT for generating PDFs, and NOT for plain-text/markdown files (read those directly).
documentation31
spec-status
Spec-driven development status and orientation. Use when checking overall project state, viewing a specific task with its linked req/epic, listing tasks by status, running a quality audit for orphans/cycles/missing fields, or for a pipeline overview when unsure which spec sub-skill to use. NOT for mutating state — read-only; use spec-done or spec-work for state changes.
development31
spec-new
Create a single TASK or REQ file from a template. Use for one-off artifact creation without the full planning workflow. NOT for full project bootstrap (spec-init) or multi-task planning from a requirement (spec-plan).
tools31
spec-plan
Turn a requirement or a concrete feature idea into an EPIC with vertical-slice TASKs. Use when you have a REQ file, or a feature idea already specific enough for a quick 3–5 question pass, and need an executable plan with dependencies and acceptance criteria. NOT for open-ended idea exploration — use brainstorming-ideas. NOT for capturing PRD-quality requirements — use spec-interview. NOT for implementing tasks — use spec-work.
testing31
spec-new
Create a single TASK or REQ file from a template. Use for one-off artifact creation without the full planning workflow. NOT for full project bootstrap (spec-init) or multi-task planning from a requirement (spec-plan).
tools31
spec-plan
Turn a requirement or a concrete feature idea into an EPIC with vertical-slice TASKs. Use when you have a REQ file, or a feature idea already specific enough for a quick 3–5 question pass, and need an executable plan with dependencies and acceptance criteria. NOT for open-ended idea exploration — use brainstorming-ideas. NOT for capturing PRD-quality requirements — use spec-interview. NOT for implementing tasks — use spec-work.
testing31
spec-status
Spec-driven development status and orientation. Use when checking overall project state, viewing a specific task with its linked req/epic, listing tasks by status, running a quality audit for orphans/cycles/missing fields, or for a pipeline overview when unsure which spec sub-skill to use. NOT for mutating state — read-only; use spec-done or spec-work for state changes.
development31
using-cloud-cli
Cloud CLI patterns for GCP and AWS. Use when running bq queries, gcloud commands, aws commands, or making decisions about cloud services. Covers BigQuery cost optimization and operational best practices. NOT for Terraform or Kubernetes architectural decisions (see managing-infra).
tools31
using-git-worktrees
Creates isolated git worktrees for parallel development. Use when starting feature work needing isolation or working on multiple branches simultaneously. Not for simple branch switching or basic git operations.
development31
watch-team
Monitor a Claude Code team in tmux, auto-approve prompts, and report status. Use when the user says "watch the team", "monitor agents", "auto-approve", "babysit team", or "agents are stuck". NOT for single-agent monitoring, non-tmux setups, or general process supervision.
development31
writing-typescript
Idiomatic TypeScript development. Use when writing TypeScript code, Node.js services, React apps, or discussing TS patterns. Emphasizes strict typing, composition, and modern tooling (bun/vite). NOT for Go, Python, plain HTML/CSS/JS, or server-rendered templates (use writing-web for those).
tools31
spec-work
Implement the next ready task. Use when starting a development session — selects the highest-priority ready task, plans with a specialist subagent, implements with approval at every step, verifies quality gates, and commits. One task per session. NOT for batch task execution or planning new work — use spec-plan for planning.
development31
reviewing-code
Code review covering security, quality, tests, implementation, documentation, and architecture / module-depth. Use when the user asks to review code, check changes, audit a PR or diff, find refactoring opportunities, or look for shallow modules and over-abstraction. NOT for fixing the issues found (use fixing-code) or applying refactors (use refactoring-code).
development31
spec-interview
Capture PRD-quality requirements through structured Q&A. Use when a new requirement needs deep exploration — produces a `REQ-*.md` via 8–15 targeted questions. NOT for creating tasks or implementation plans — use spec-plan for that.
testing31
improving-tests
Improve test design and coverage, including TDD/red-green-refactor guidance. Use when improving tests, refactoring tests, adding coverage, using TDD, or removing test waste. NOT for fixing production bugs (use fixing-code) or reviewing non-test code quality (use reviewing-code).
development31
testing-e2e
Playwright-based end-to-end browser testing of user flows. Use when running existing Playwright tests, generating browser checks, recording a visible session, or verifying a user journey end-to-end. NOT for unit tests, API-only tests, or logic tests where curl or JSDOM suffices — use improving-tests or fixing-code instead.
development31
spec-work
Implement the next ready task. Use when starting a development session — selects the highest-priority ready task, plans with a specialist subagent, implements with approval at every step, verifies quality gates, and commits. One task per session. NOT for batch task execution or planning new work — use spec-plan for planning.
development31
context7-cli
ctx7 (Context7) CLI mechanics for querying versioned library documentation. Use when the user mentions "ctx7" or "context7", passes a `/org/project` library ID, or another skill needs the exact ctx7 command workflow. NOT the docs-lookup decision flow or web fallback — that is looking-up-docs.
tools31
managing-infra
Infrastructure patterns for Kubernetes, Terraform, Helm, Kustomize, and GitHub Actions. Use when making K8s architectural decisions, choosing between Helm vs Kustomize, structuring Terraform modules, writing CI/CD workflows, or applying security best practices. NOT for cloud CLI commands (see using-cloud-cli) or deploy validation and apply workflows (see deploying-infra).
tools31
spec-interview
Capture PRD-quality requirements through structured Q&A. Use when a new requirement needs deep exploration — produces a `REQ-*.md` via 8–15 targeted questions. NOT for creating tasks or implementation plans — use spec-plan for that.
testing31
improving-codebase-architecture
Find deepening opportunities in a codebase, informed by the domain language in CONTEXT.md and the decisions in docs/adr/. Use when the user wants to improve architecture, find refactoring opportunities, consolidate tightly-coupled modules, or make a codebase more testable and AI-navigable. NOT for line-level cleanup (use reviewing-code) or batch edits (use refactoring-code).
development31
testing-e2e
Playwright-based end-to-end browser testing of user flows. Use when running existing Playwright tests, generating browser checks, recording a visible session, or verifying a user journey end-to-end. NOT for unit tests, API-only tests, or logic tests where curl or JSDOM suffices — use improving-tests or fixing-code instead.
development31
spec-work
Implement the next ready task. Use when starting a development session — selects the highest-priority ready task, plans with a specialist subagent, implements with approval at every step, verifies quality gates, and commits. One task per session. NOT for batch task execution or planning new work — use spec-plan for planning.
development31
context7-cli
ctx7 (Context7) CLI mechanics for querying versioned library documentation. Use when the user mentions "ctx7" or "context7", passes a `/org/project` library ID, or another skill needs the exact ctx7 command workflow. NOT the docs-lookup decision flow or web fallback — that is looking-up-docs.
tools31
spec-status
Spec-driven development status and orientation. Use when checking overall project state, viewing a specific task with its linked req/epic, listing tasks by status, running a quality audit for orphans/cycles/missing fields, or for a pipeline overview when unsure which spec sub-skill to use. NOT for mutating state — read-only; use spec-done or spec-work for state changes.
development31
spec-new
Create a single TASK or REQ file from a template. Use for one-off artifact creation without the full planning workflow. NOT for full project bootstrap (spec-init) or multi-task planning from a requirement (spec-plan).
tools31
spec-plan
Turn a requirement or a concrete feature idea into an EPIC with vertical-slice TASKs. Use when you have a REQ file, or a feature idea already specific enough for a quick 3–5 question pass, and need an executable plan with dependencies and acceptance criteria. NOT for open-ended idea exploration — use brainstorming-ideas. NOT for capturing PRD-quality requirements — use spec-interview. NOT for implementing tasks — use spec-work.
testing31
testing-e2e
Playwright-based end-to-end browser testing of user flows. Use when running existing Playwright tests, generating browser checks, recording a visible session, or verifying a user journey end-to-end. NOT for unit tests, API-only tests, or logic tests where curl or JSDOM suffices — use improving-tests or fixing-code instead.
development31
using-cloud-cli
Cloud CLI patterns for GCP and AWS. Use when running bq queries, gcloud commands, aws commands, or making decisions about cloud services. Covers BigQuery cost optimization and operational best practices. NOT for Terraform or Kubernetes architectural decisions (see managing-infra).
tools31
watch-team
Monitor a Claude Code team in tmux, auto-approve prompts, and report status. Use when the user says "watch the team", "monitor agents", "auto-approve", "babysit team", or "agents are stuck". NOT for single-agent monitoring, non-tmux setups, or general process supervision.
development31
analyzing-usage
Analyze AI coding agent usage, cost, efficiency, and burn rate — Claude Code (ccusage), Pi/pi-agent (ccusage-pi), or Codex CLI (ccusage-codex). Use when user says "usage", "cost", "spending", "tokens", "analyze usage", "how much did I spend", "usage report", "budget", "burn rate", "efficiency", "cache hits", "ccusage", "ccw", "ccp". NOT for general shell scripting, non-AI-agent cost analysis, or cloud infrastructure billing (use using-cloud-cli).
tools31
mem-history
Query project history, past decisions, and known gotchas from claude-mem observations. Use when user asks "last session", "did we already", "what did we decide", "project history", "timeline", or "what happened with".
development31
reviewing-cc-config
Review Claude Code configuration for context efficiency, signal density, and anti-patterns. Use when user says "review config", "review setup", "check configuration", "review cc config", "context review", "config review", "review my setup", "review skills", "review agents", "review hooks", or wants feedback on their Claude Code configuration quality. NOT for editing config files — review only; user applies fixes unless --fix is passed. NOT for applying config changes (use `evolving-config`).
development31
parsing-documents
Extract structured data from PDF documents — text, tables, forms, and metadata. Use when reading or extracting content from a `.pdf` file, parsing invoices/reports/scanned documents, or converting PDF data to JSON/CSV. NOT for generating PDFs, and NOT for plain-text/markdown files (read those directly).
documentation31
spec-done
Mark a task complete with evidence. Use when finishing a task, discovering which in-progress tasks look done from git history, or verifying quality gates before closing out. Handles follow-up task creation and durable learnings. NOT for reporting progress (spec-status).
testing31
spec-init
Initialize a `.spec/` project or extract requirements from a document. Use when there is no `.spec/` directory yet, or to add requirements from an existing design doc. NOT for one-off task/req creation (spec-new) or deep PRD-quality requirement capture (spec-interview).
testing31
managing-infra
Infrastructure patterns for Kubernetes, Terraform, Helm, Kustomize, and GitHub Actions. Use when making K8s architectural decisions, choosing between Helm vs Kustomize, structuring Terraform modules, writing CI/CD workflows, or applying security best practices. NOT for cloud CLI commands (see using-cloud-cli) or deploy validation and apply workflows (see deploying-infra).
tools31
testing-e2e
Playwright-based end-to-end browser testing of user flows. Use when running existing Playwright tests, generating browser checks, recording a visible session, or verifying a user journey end-to-end. NOT for unit tests, API-only tests, or logic tests where curl or JSDOM suffices — use improving-tests or fixing-code instead.
development31
spec-new
Create a single TASK or REQ file from a template. Use for one-off artifact creation without the full planning workflow. NOT for full project bootstrap (spec-init) or multi-task planning from a requirement (spec-plan).
tools31
spec-work
Implement the next ready task. Use when starting a development session — selects the highest-priority ready task, plans with a specialist subagent, implements with approval at every step, verifies quality gates, and commits. One task per session. NOT for batch task execution or planning new work — use spec-plan for planning.
development31
spec-status
Spec-driven development status and orientation. Use when checking overall project state, viewing a specific task with its linked req/epic, listing tasks by status, running a quality audit for orphans/cycles/missing fields, or for a pipeline overview when unsure which spec sub-skill to use. NOT for mutating state — read-only; use spec-done or spec-work for state changes.
development31
spec-plan
Turn a requirement or a concrete feature idea into an EPIC with vertical-slice TASKs. Use when you have a REQ file, or a feature idea already specific enough for a quick 3–5 question pass, and need an executable plan with dependencies and acceptance criteria. NOT for open-ended idea exploration — use brainstorming-ideas. NOT for capturing PRD-quality requirements — use spec-interview. NOT for implementing tasks — use spec-work.
testing31
parsing-documents
Extract structured data from PDF documents — text, tables, forms, and metadata. Use when reading or extracting content from a `.pdf` file, parsing invoices/reports/scanned documents, or converting PDF data to JSON/CSV. NOT for generating PDFs, and NOT for plain-text/markdown files (read those directly).
documentation31
improving-codebase-architecture
Find deepening opportunities in a codebase, informed by the domain language in CONTEXT.md and the decisions in docs/adr/. Use when the user wants to improve architecture, find refactoring opportunities, consolidate tightly-coupled modules, or make a codebase more testable and AI-navigable. NOT for line-level cleanup (use reviewing-code) or batch edits (use refactoring-code).
development31
improving-codebase-architecture
Find deepening opportunities in a codebase, informed by the domain language in CONTEXT.md and the decisions in docs/adr/. Use when the user wants to improve architecture, find refactoring opportunities, consolidate tightly-coupled modules, or make a codebase more testable and AI-navigable. NOT for line-level cleanup (use reviewing-code) or batch edits (use refactoring-code).
development31
brainstorming-ideas
Brainstorm ideas and stress-test draft plans before coding. Use when brainstorming, exploring approaches, designing a feature/API/flow, grilling a plan, challenging assumptions, or resolving terminology that blocks the design. NOT for implementation task breakdown; use the spec-plan skill. NOT for general documentation updates; use documenting-code or learning-patterns.
development31
looking-up-docs
Find current, factual library/API/framework documentation through a tool-fallback chain. Use when the user says "look up docs", "how to use", "API for", "syntax for", "examples of", "show me the docs", or wants the latest/current/actual behavior of a library, framework, CLI, or API. NOT for comparisons, best-practice surveys, or recent ecosystem news — use researching-web. NOT for raw ctx7 CLI mechanics — that is context7-cli.
tools31
writing-web
Simple web development with HTML, CSS, JS, and HTMX. Use when working with .html, .css, or .htmx files, web templates, stylesheets, or vanilla JS scripts. NOT for React/Vue/Angular (use writing-typescript) or Node.js backends.
development30
spec-core
Spec-driven development orientation and quick reference. Use when starting spec-driven development, unsure which sub-skill to use, or wanting a pipeline overview and current project state. NOT for executing tasks (spec-work) or marking them done (spec-done).
development30
watch-team
Monitor a Claude Code team in tmux, auto-approve prompts, and report status. Use when the user says "watch the team", "monitor agents", "auto-approve", "babysit team", or "agents are stuck". NOT for single-agent monitoring, non-tmux setups, or general process supervision.
development30
spec-plan
Turn a requirement or a concrete feature idea into an EPIC with vertical-slice TASKs. Use when you have a REQ file, or a feature idea already specific enough for a quick 3–5 question pass, and need an executable plan with dependencies and acceptance criteria. NOT for open-ended idea exploration — use brainstorming-ideas. NOT for capturing PRD-quality requirements — use spec-interview. NOT for implementing tasks — use spec-work.
testing30
improving-codebase-architecture
Find deepening opportunities in a codebase, informed by the domain language in CONTEXT.md and the decisions in docs/adr/. Use when the user wants to improve architecture, find refactoring opportunities, consolidate tightly-coupled modules, or make a codebase more testable and AI-navigable. NOT for line-level cleanup (use reviewing-code) or batch edits (use refactoring-code).
development30
searching-code
Intelligent codebase search with AST-first local search and zoom-out mapping. Use when user asks "how does X work", "trace flow", "find all implementations", "understand codebase", "zoom out", "map this area", structural code-pattern search, or cross-file exploration in large repos. Try ast-grep before rg for code-shape queries; use WarpGrep for semantic flow.
development30
searching-code
Intelligent codebase search with AST-first local search and zoom-out mapping. Use when user asks "how does X work", "trace flow", "find all implementations", "understand codebase", "zoom out", "map this area", structural code-pattern search, or cross-file exploration in large repos. Try ast-grep before rg for code-shape queries; use WarpGrep for semantic flow.
development30
searching-code
Intelligent codebase search with AST-first local search and zoom-out mapping. Use when user asks "how does X work", "trace flow", "find all implementations", "understand codebase", "zoom out", "map this area", structural code-pattern search, or cross-file exploration in large repos. Try ast-grep before rg for code-shape queries; use WarpGrep for semantic flow.
development30
searching-code
Intelligent codebase search with AST-first local search and zoom-out mapping. Use when user asks "how does X work", "trace flow", "find all implementations", "understand codebase", "zoom out", "map this area", structural code-pattern search, or cross-file exploration in large repos. Try ast-grep before rg for code-shape queries; use WarpGrep for semantic flow.
development30
spec-interview
Capture PRD-quality requirements through structured Q&A. Use when a new requirement needs deep exploration — produces a `REQ-*.md` via 8–15 targeted questions. NOT for creating tasks or implementation plans — use spec-plan for that.
testing30
learning-patterns
Extract durable learnings from a session and propose project customizations — agent-instructions file, CONTEXT.md, ADRs, project skills, hooks. Use when the user says "learn", "extract learnings", "what did we learn", "save learnings", "adapt config", "capture domain language", or wants to encode session patterns durably. NOT for documentation edits (use documenting-code) or committing changes (use committing-code).
development30
improving-codebase-architecture
Find deepening opportunities in a codebase, informed by the domain language in CONTEXT.md and the decisions in docs/adr/. Use when the user wants to improve architecture, find refactoring opportunities, consolidate tightly-coupled modules, or make a codebase more testable and AI-navigable. NOT for line-level cleanup (use reviewing-code) or batch edits (use refactoring-code).
development30
sequential-thinking
Structured stepwise reasoning with explicit revisions and branches. Use when the user says "think step by step", "sequential thinking", "plan this out", "reason through this", "branch this idea", or when tackling a hard multi-step problem (architecture decisions, ambiguous bugs, multi-constraint tradeoffs, plans that may need revision). NOT for trivial lookups, single-tool fetches, or tasks the model can answer directly without planning.
tools30
watch-team
Monitor a Claude Code team in tmux, auto-approve prompts, and report status. Use when the user says "watch the team", "monitor agents", "auto-approve", "babysit team", or "agents are stuck". NOT for single-agent monitoring, non-tmux setups, or general process supervision.
development30
spec-new
Create a single TASK or REQ file from a template. Use for one-off artifact creation without the full planning workflow. NOT for full project bootstrap (spec-init) or multi-task planning from a requirement (spec-plan).
tools30
ccgram-messaging
Inter-agent messaging via ccgram swarm. Use when communicating with other agents in the same tmux session — send messages, check inbox, discover peers, broadcast status, reply to requests, or spawn new agents. Activates on: peer messages, inbox, swarm, ccgram, broadcast, agent collaboration, ask another agent.
testing30
spec-done
Mark a task complete with evidence. Use when finishing a task, discovering which in-progress tasks look done from git history, or verifying quality gates before closing out. Handles follow-up task creation and durable learnings. NOT for reporting progress (spec-status).
testing30
searching-code
Intelligent codebase search with AST-first local search and zoom-out mapping. Use when user asks "how does X work", "trace flow", "find all implementations", "understand codebase", "zoom out", "map this area", structural code-pattern search, or cross-file exploration in large repos. Try ast-grep before rg for code-shape queries; use WarpGrep for semantic flow.
development30
playwright-skill
Playwright primitives for real-browser automation — dev-server detection, a Node.js script runner, and helpers for clicks, form fills, screenshots, multi-viewport, custom HTTP headers. Use when a task needs an actual browser (rendered DOM, visual checks, multi-page flows, cross-browser behavior). Not for API tests or logic tests where curl or JSDOM is cheaper.
tools30
testing-e2e
Playwright-based end-to-end browser testing of user flows. Use when running existing Playwright tests, generating browser checks, recording a visible session, or verifying a user journey end-to-end. NOT for unit tests, API-only tests, or logic tests where curl or JSDOM suffices — use improving-tests or fixing-code instead.
development30
spec-work
Implement the next ready task. Use when starting a development session — selects the highest-priority ready task, plans with a specialist subagent, implements with approval at every step, verifies quality gates, and commits. One task per session. NOT for batch task execution or planning new work — use spec-plan for planning.
development30
spec-status
Spec-driven development status and orientation. Use when checking overall project state, viewing a specific task with its linked req/epic, listing tasks by status, running a quality audit for orphans/cycles/missing fields, or for a pipeline overview when unsure which spec sub-skill to use. NOT for mutating state — read-only; use spec-done or spec-work for state changes.
development30
ccgram-messaging
Inter-agent messaging via ccgram swarm. Use when communicating with other agents in the same tmux session — send messages, check inbox, discover peers, broadcast status, reply to requests, or spawn new agents. Activates on: peer messages, inbox, swarm, ccgram, broadcast, agent collaboration, ask another agent.
testing30
watch-team
Monitor a Claude Code team in tmux, auto-approve prompts, and report status. Use when the user says "watch the team", "monitor agents", "auto-approve", "babysit team", or "agents are stuck". NOT for single-agent monitoring, non-tmux setups, or general process supervision.
development30
spec-done
Mark a task complete with evidence. Use when finishing a task, discovering which in-progress tasks look done from git history, or verifying quality gates before closing out. Handles follow-up task creation and durable learnings. NOT for reporting progress (spec-status).
testing30
spec-init
Initialize a `.spec/` project or extract requirements from a document. Use when there is no `.spec/` directory yet, or to add requirements from an existing design doc. NOT for one-off task/req creation (spec-new) or deep PRD-quality requirement capture (spec-interview).
testing30
spec-new
Create a single TASK or REQ file from a template. Use for one-off artifact creation without the full planning workflow. NOT for full project bootstrap (spec-init) or multi-task planning from a requirement (spec-plan).
tools30
spec-interview
Capture PRD-quality requirements through structured Q&A. Use when a new requirement needs deep exploration — produces a `REQ-*.md` via 8–15 targeted questions. NOT for creating tasks or implementation plans — use spec-plan for that.
testing30
spec-plan
Turn a requirement or a concrete feature idea into an EPIC with vertical-slice TASKs. Use when you have a REQ file, or a feature idea already specific enough for a quick 3–5 question pass, and need an executable plan with dependencies and acceptance criteria. NOT for open-ended idea exploration — use brainstorming-ideas. NOT for capturing PRD-quality requirements — use spec-interview. NOT for implementing tasks — use spec-work.
testing30
spec-status
Spec-driven development status and orientation. Use when checking overall project state, viewing a specific task with its linked req/epic, listing tasks by status, running a quality audit for orphans/cycles/missing fields, or for a pipeline overview when unsure which spec sub-skill to use. NOT for mutating state — read-only; use spec-done or spec-work for state changes.
development30
spec-work
Implement the next ready task. Use when starting a development session — selects the highest-priority ready task, plans with a specialist subagent, implements with approval at every step, verifies quality gates, and commits. One task per session. NOT for batch task execution or planning new work — use spec-plan for planning.
development30
testing-e2e
Playwright-based end-to-end browser testing of user flows. Use when running existing Playwright tests, generating browser checks, recording a visible session, or verifying a user journey end-to-end. NOT for unit tests, API-only tests, or logic tests where curl or JSDOM suffices — use improving-tests or fixing-code instead.
development30
playwright-skill
Playwright primitives for real-browser automation — dev-server detection, a Node.js script runner, and helpers for clicks, form fills, screenshots, multi-viewport, custom HTTP headers. Use when a task needs an actual browser (rendered DOM, visual checks, multi-page flows, cross-browser behavior). Not for API tests or logic tests where curl or JSDOM is cheaper.
tools30
spec-init
Initialize a `.spec/` project or extract requirements from a document. Use when there is no `.spec/` directory yet, or to add requirements from an existing design doc. NOT for one-off task/req creation (spec-new) or deep PRD-quality requirement capture (spec-interview).
testing30
spec-core
Spec-driven development orientation and quick reference. Use when starting spec-driven development, unsure which sub-skill to use, or wanting a pipeline overview and current project state. NOT for executing tasks (spec-work) or marking them done (spec-done).
development27
playwright-skill
Playwright primitives for real-browser automation — dev-server detection, a Node.js script runner, and helpers for clicks, form fills, screenshots, multi-viewport, custom HTTP headers. Use when a task needs an actual browser (rendered DOM, visual checks, multi-page flows, cross-browser behavior). Not for API tests or logic tests where curl or JSDOM is cheaper.
tools27
spec-init
Initialize a `.spec/` project or extract requirements from a document. Use when there is no `.spec/` directory yet, or to add requirements from an existing design doc. NOT for one-off task/req creation (spec-new) or deep PRD-quality requirement capture (spec-interview).
testing27
grill-me
Interview the user relentlessly about a plan or design until reaching shared understanding, resolving each branch of the decision tree. Use when user says "grill me", wants to stress-test a single plan, or asks to be challenged on a specific design. NOT for full ideation/feature design (use brainstorming-ideas) or thesis-vs-antithesis debates (use debating-ideas).
testing27
learning-patterns
Extract durable learnings from a session and propose project customizations — agent-instructions file, CONTEXT.md, ADRs, project skills, hooks. Use when the user says "learn", "extract learnings", "what did we learn", "save learnings", "adapt config", "capture domain language", or wants to encode session patterns durably. NOT for documentation edits (use documenting-code) or committing changes (use committing-code).
development27
sequential-thinking
Structured stepwise reasoning with explicit revisions and branches. Use when the user says "think step by step", "sequential thinking", "plan this out", "reason through this", "branch this idea", or when tackling a hard multi-step problem (architecture decisions, ambiguous bugs, multi-constraint tradeoffs, plans that may need revision). NOT for trivial lookups, single-tool fetches, or tasks the model can answer directly without planning.
tools27
fixing-code
Fix code problems with disciplined diagnosis — run checks, build a repro for bugs, rank falsifiable hypotheses, fix one issue at a time, and verify until clean. Use when fixing, debugging, diagnosing, or resolving lint/test/build failures.
development27
spec-done
Mark a task complete with evidence. Use when finishing a task, discovering which in-progress tasks look done from git history, or verifying quality gates before closing out. Handles follow-up task creation and durable learnings. NOT for reporting progress (spec-status).
testing27
spec-init
Initialize a `.spec/` project or extract requirements from a document. Use when there is no `.spec/` directory yet, or to add requirements from an existing design doc. NOT for one-off task/req creation (spec-new) or deep PRD-quality requirement capture (spec-interview).
testing27
spec-core
Spec-driven development orientation and quick reference. Use when starting spec-driven development, unsure which sub-skill to use, or wanting a pipeline overview and current project state. NOT for executing tasks (spec-work) or marking them done (spec-done).
development27
playwright-skill
Playwright primitives for real-browser automation — dev-server detection, a Node.js script runner, and helpers for clicks, form fills, screenshots, multi-viewport, custom HTTP headers. Use when a task needs an actual browser (rendered DOM, visual checks, multi-page flows, cross-browser behavior). Not for API tests or logic tests where curl or JSDOM is cheaper.
tools27
improve-codebase-architecture
Find deepening opportunities in a codebase, informed by the domain language in CONTEXT.md and the decisions in docs/adr/. Use when the user wants to improve architecture, find refactoring opportunities, consolidate tightly-coupled modules, or make a codebase more testable and AI-navigable. NOT for line-level cleanup (use reviewing-code) or batch edits (use refactoring-code).
development27
ccgram-messaging
Inter-agent messaging via ccgram swarm. Use when communicating with other agents in the same tmux session — send messages, check inbox, discover peers, broadcast status, reply to requests, or spawn new agents. Activates on: peer messages, inbox, swarm, ccgram, broadcast, agent collaboration, ask another agent.
testing27
analyzing-usage
Analyze pi-agent session usage, token cost, and efficiency using ccusage-pi. Use when user says "usage", "cost", "tokens", "how much did I spend", "pi usage", "session cost", "ccusage-pi". NOT for cloud billing or non-pi-agent usage analysis.
testing27
spec-core
Spec-driven development orientation and quick reference. Use when starting spec-driven development, unsure which sub-skill to use, or wanting a pipeline overview and current project state. NOT for executing tasks (spec-work) or marking them done (spec-done).
development27
spec-done
Mark a task complete with evidence. Use when finishing a task, discovering which in-progress tasks look done from git history, or verifying quality gates before closing out. Handles follow-up task creation and durable learnings. NOT for reporting progress (spec-status).
testing27
spec-init
Initialize a `.spec/` project or extract requirements from a document. Use when there is no `.spec/` directory yet, or to add requirements from an existing design doc. NOT for one-off task/req creation (spec-new) or deep PRD-quality requirement capture (spec-interview).
testing27
spec-done
Mark a task complete with evidence. Use when finishing a task, discovering which in-progress tasks look done from git history, or verifying quality gates before closing out. Handles follow-up task creation and durable learnings. NOT for reporting progress (spec-status).
testing27
debating-ideas
Dialectic thinking for code and architecture decisions — spawn thesis and antithesis agents, verify claims against the codebase, then synthesize. Use when user says "debate", "argue both sides", "devil's advocate", "pros and cons of approach", or wants a design decision stress-tested against actual code. For conceptual or logical claims without a codebase, use thinking-tools:dialectic instead.
tools27
committing-code
Smart git commits with logical grouping. Use when user says "commit", "commit changes", "save changes", "create commit", "bundle commits", "git commit", or wants to commit their work.
development27
grill-me
Interview the user relentlessly about a plan or design until reaching shared understanding, resolving each branch of the decision tree. Use when user says "grill me", wants to stress-test a single plan, or asks to be challenged on a specific design. NOT for full ideation/feature design (use brainstorming-ideas) or thesis-vs-antithesis debates (use debating-ideas).
testing27
grill-me
Interview the user relentlessly about a plan or design until reaching shared understanding, resolving each branch of the decision tree. Use when user says "grill me", wants to stress-test a single plan, or asks to be challenged on a specific design. NOT for full ideation/feature design (use brainstorming-ideas) or thesis-vs-antithesis debates (use debating-ideas).
testing27
searching-code
Intelligent codebase search and zoom-out mapping via WarpGrep. Use when user asks "how does X work", "trace flow", "find all implementations", "understand codebase", "zoom out", "map this area", or needs cross-file exploration in large repos (1000+ files).
development27
sequential-thinking
Structured stepwise reasoning with explicit revisions and branches. Use when the user says "think step by step", "sequential thinking", "plan this out", "reason through this", "branch this idea", or when tackling a hard multi-step problem (architecture decisions, ambiguous bugs, multi-constraint tradeoffs, plans that may need revision). NOT for trivial lookups, single-tool fetches, or tasks the model can answer directly without planning.
tools27
learning-patterns
Extract durable learnings from a session and propose project customizations — agent-instructions file, CONTEXT.md, ADRs, project skills, hooks. Use when the user says "learn", "extract learnings", "what did we learn", "save learnings", "adapt config", "capture domain language", or wants to encode session patterns durably. NOT for documentation edits (use documenting-code) or committing changes (use committing-code).
development27
learning-patterns
Extract durable learnings from a session and propose project customizations — agent-instructions file, CONTEXT.md, ADRs, project skills, hooks. Use when the user says "learn", "extract learnings", "what did we learn", "save learnings", "adapt config", "capture domain language", or wants to encode session patterns durably. NOT for documentation edits (use documenting-code) or committing changes (use committing-code).
development27
ccgram-messaging
Inter-agent messaging via ccgram swarm. Use when communicating with other agents in the same tmux session — send messages, check inbox, discover peers, broadcast status, reply to requests, or spawn new agents. Activates on: peer messages, inbox, swarm, ccgram, broadcast, agent collaboration, ask another agent.
testing27
analyzing-usage
Analyze Codex CLI session usage, token cost, and efficiency using ccusage-codex. Use when user says "usage", "cost", "tokens", "how much did I spend", "codex usage", "session cost", "ccusage-codex". NOT for cloud billing or non-codex usage analysis.
tools27
searching-code
Intelligent codebase search and zoom-out mapping via WarpGrep. Use when user asks "how does X work", "trace flow", "find all implementations", "understand codebase", "zoom out", "map this area", or needs cross-file exploration in large repos (1000+ files).
development27
spec-core
Spec-driven development orientation and quick reference. Use when starting spec-driven development, unsure which sub-skill to use, or wanting a pipeline overview and current project state. NOT for executing tasks (spec-work) or marking them done (spec-done).
development27
spec-done
Mark a task complete with evidence. Use when finishing a task, discovering which in-progress tasks look done from git history, or verifying quality gates before closing out. Handles follow-up task creation and durable learnings. NOT for reporting progress (spec-status).
testing27
ccgram-messaging
Inter-agent messaging via ccgram swarm. Use when communicating with other agents in the same tmux session — send messages, check inbox, discover peers, broadcast status, reply to requests, or spawn new agents. Activates on: peer messages, inbox, swarm, ccgram, broadcast, agent collaboration, ask another agent.
testing27
grill-me
Interview the user relentlessly about a plan or design until reaching shared understanding, resolving each branch of the decision tree. Use when user says "grill me", wants to stress-test a single plan, or asks to be challenged on a specific design. NOT for full ideation/feature design (use brainstorming-ideas) or thesis-vs-antithesis debates (use debating-ideas).
testing27
improve-codebase-architecture
Find deepening opportunities in a codebase, informed by the domain language in CONTEXT.md and the decisions in docs/adr/. Use when the user wants to improve architecture, find refactoring opportunities, consolidate tightly-coupled modules, or make a codebase more testable and AI-navigable. NOT for line-level cleanup (use reviewing-code) or batch edits (use refactoring-code).
development27
learning-patterns
Extract durable learnings from a session and propose project customizations — agent-instructions file, CONTEXT.md, ADRs, project skills, hooks. Use when the user says "learn", "extract learnings", "what did we learn", "save learnings", "adapt config", "capture domain language", or wants to encode session patterns durably. NOT for documentation edits (use documenting-code) or committing changes (use committing-code).
development27
sequential-thinking
Structured stepwise reasoning with explicit revisions and branches. Use when the user says "think step by step", "sequential thinking", "plan this out", "reason through this", "branch this idea", or when tackling a hard multi-step problem (architecture decisions, ambiguous bugs, multi-constraint tradeoffs, plans that may need revision). NOT for trivial lookups, single-tool fetches, or tasks the model can answer directly without planning.
tools27
committing-code
Smart git commits with logical grouping. Use when user says "commit", "commit changes", "save changes", "create commit", "bundle commits", "git commit", or wants to commit their work.
development27
spec-init
Initialize a `.spec/` project or extract requirements from a document. Use when there is no `.spec/` directory yet, or to add requirements from an existing design doc. NOT for one-off task/req creation (spec-new) or deep PRD-quality requirement capture (spec-interview).
testing27
improve-codebase-architecture
Find deepening opportunities in a codebase, informed by the domain language in CONTEXT.md and the decisions in docs/adr/. Use when the user wants to improve architecture, find refactoring opportunities, consolidate tightly-coupled modules, or make a codebase more testable and AI-navigable. NOT for line-level cleanup (use reviewing-code) or batch edits (use refactoring-code).
development27
grill-me
Interview the user relentlessly about a plan or design until reaching shared understanding, resolving each branch of the decision tree. Use when user says "grill me", wants to stress-test a single plan, or asks to be challenged on a specific design. NOT for full ideation/feature design (use brainstorming-ideas) or thesis-vs-antithesis debates (use debating-ideas).
testing27
sequential-thinking
Structured stepwise reasoning with explicit revisions and branches. Use when the user says "think step by step", "sequential thinking", "plan this out", "reason through this", "branch this idea", or when tackling a hard multi-step problem (architecture decisions, ambiguous bugs, multi-constraint tradeoffs, plans that may need revision). NOT for trivial lookups, single-tool fetches, or tasks the model can answer directly without planning.
tools27
debating-ideas
Dialectic thinking for code and architecture decisions — spawn thesis and antithesis agents, verify claims against the codebase, then synthesize. Use when user says "debate", "argue both sides", "devil's advocate", "pros and cons of approach", or wants a design decision stress-tested against actual code. For conceptual or logical claims without a codebase, use thinking-tools:dialectic instead.
tools27
improve-codebase-architecture
Find deepening opportunities in a codebase, informed by the domain language in CONTEXT.md and the decisions in docs/adr/. Use when the user wants to improve architecture, find refactoring opportunities, consolidate tightly-coupled modules, or make a codebase more testable and AI-navigable. NOT for line-level cleanup (use reviewing-code) or batch edits (use refactoring-code).
development27
ccgram-messaging
Inter-agent messaging via ccgram swarm. Use when communicating with other agents in the same tmux session — send messages, check inbox, discover peers, broadcast status, reply to requests, or spawn new agents. Activates on: peer messages, inbox, swarm, ccgram, broadcast, agent collaboration, ask another agent.
testing27
improve-codebase-architecture
Find deepening opportunities in a codebase, informed by the domain language in CONTEXT.md and the decisions in docs/adr/. Use when the user wants to improve architecture, find refactoring opportunities, consolidate tightly-coupled modules, or make a codebase more testable and AI-navigable. NOT for line-level cleanup (use reviewing-code) or batch edits (use refactoring-code).
development27
ccgram-messaging
Inter-agent messaging via ccgram swarm. Use when communicating with other agents in the same tmux session — send messages, check inbox, discover peers, broadcast status, reply to requests, or spawn new agents. Activates on: peer messages, inbox, swarm, ccgram, broadcast, agent collaboration, ask another agent.
testing27
improving-tests
Improve test design and coverage, including TDD/red-green-refactor guidance. Use when improving tests, refactoring tests, adding coverage, using TDD, or removing test waste.
development21
searching-code
Intelligent codebase search and zoom-out mapping using Pi tools. Use when the user asks how code works, to trace a flow, find implementations, zoom out, or map an area of a large repo.
tools21
fixing-code
Fix code problems with disciplined diagnosis — run checks, build a repro for bugs, rank falsifiable hypotheses, fix one issue at a time, and verify until clean. Use when fixing, debugging, diagnosing, or resolving lint/test/build failures.
development21
exploring-repos
Explore public GitHub repositories via DeepWiki AI-generated documentation. Use for understanding architecture, patterns, design decisions, and code organization of popular open-source projects. Use when user asks "how does X repo work", "explain architecture of Y", "what patterns does Z use", "explore repo", "deepwiki", or needs codebase-level understanding beyond API docs.
development21
writing-web
Simple web development with HTML, CSS, JS, and HTMX. Auto-activates when working with .html, .css, or .htmx files, web templates, stylesheets, or vanilla JS scripts. NOT for React/Vue/Angular (use writing-typescript) or Node.js backends.
development21
testing-e2e
Sequential E2E workflow — identify test targets, generate Playwright test scripts written to /tmp, run them with node, capture failures, fix and re-run until passing. Supports TypeScript tests and Go/HTMX applications.
development21
writing-web
Simple web development with HTML, CSS, JS, and HTMX. Auto-activates when working with .html, .css, or .htmx files, web templates, stylesheets, or vanilla JS scripts. NOT for React/Vue/Angular (use writing-typescript) or Node.js backends.
development21
improving-tests
Improve test design and coverage, including TDD/red-green-refactor guidance. Use when improving tests, refactoring tests, adding coverage, using TDD, or removing test waste.
development21
deploying-infra
Sequential infrastructure deployment — detect infra type (Kubernetes, Terraform, Helm, Kustomize, GitHub Actions, Docker), validate configs, dry-run, show diff, apply only after user confirmation, and verify post-deploy health. Includes safety checks for destructive operations.
testing21
writing-typescript
Idiomatic TypeScript development. Use when writing TypeScript code, Node.js services, React apps, or discussing TS patterns. Emphasizes strict typing, composition, and modern tooling (bun/vite).
tools21
improve-codebase-architecture
Find deepening opportunities in a codebase, informed by the domain language in CONTEXT.md and the decisions in docs/adr/. Use when the user wants to improve architecture, find refactoring opportunities, consolidate tightly-coupled modules, or make a codebase more testable and AI-navigable. NOT for line-level cleanup (use reviewing-code) or batch edits (use refactoring-code).
development21
deploying-infra
Sequential infrastructure deployment — detect infra type (Kubernetes, Terraform, Helm, Kustomize, GitHub Actions, Docker), validate configs, dry-run, show diff, apply only after user confirmation, and verify post-deploy health. Includes safety checks for destructive operations.
testing21
linting-instructions
Lint plugin agent/skill prompts against rules derived from Anthropic model cards (Opus 4.6, Sonnet 4.6). Use when authoring or reviewing skills and agents — "lint instructions", "audit prompts", "model card rules".
tools21
writing-python
Idiomatic Python 3.12+ development. Use when writing Python code, CLI tools, scripts, or services. Emphasizes stdlib, type hints, uv/ruff/pyright toolchain, and minimal dependencies.
tools21
playwright-skill
Internal Playwright automation library. Use when loaded by testing-e2e or playwright-tester for dev server detection, script execution, and browser automation primitives. Not directly invoked by users.
tools21
using-cloud-cli
Cloud CLI patterns for GCP and AWS. Use when running bq queries, gcloud commands, aws commands, or making decisions about cloud services. Covers BigQuery cost optimization and operational best practices.
tools21
using-cloud-cli
Cloud CLI patterns for GCP and AWS. Use when running bq queries, gcloud commands, aws commands, or making decisions about cloud services. Covers BigQuery cost optimization and operational best practices.
tools21
writing-python
Idiomatic Python 3.12+ development. Use when writing Python code, CLI tools, scripts, or services. Emphasizes stdlib, type hints, uv/ruff/pyright toolchain, and minimal dependencies.
tools21
using-cloud-cli
Cloud CLI patterns for GCP and AWS. Use when running bq queries, gcloud commands, aws commands, or making decisions about cloud services. Covers BigQuery cost optimization and operational best practices.
tools21
using-modern-cli
Prefer modern CLI tools for better performance: rg (ripgrep) instead of grep for text searching, fd instead of find for file discovery, bat instead of cat for viewing files, sd instead of sed for text replacement, eza instead of ls for directory listing, dust instead of du, procs instead of ps. Use when writing bash scripts, optimizing command chains, working with file searches, or replacing grep/find/cat in workflows. These tools are faster, respect .gitignore, and have better output formatting.
tools21
smart-explore
Token-efficient code navigation using local Pi tools. Use when you need a fast outline of files, symbols, imports, or call sites without loading whole files.
tools21
reviewing-code
Sequential code review for security, quality, tests, and architecture. Use when reviewing code, checking changes, reviewing PRs, or looking for deep-module/refactoring opportunities.
development21
fixing-code
Fix code problems with disciplined diagnosis — run checks, build a repro for bugs, rank falsifiable hypotheses, fix one issue at a time, and verify until clean. Use when fixing, debugging, diagnosing, or resolving lint/test/build failures.
development21
documenting-code
Update project documentation based on code changes. Use when the user asks to update docs, document behavior, add README content, or align docs with recent implementation changes.
development21
ccgram-messaging
Inter-agent messaging via ccgram swarm. Use when communicating with other agents in the same tmux session — send messages, check inbox, discover peers, broadcast status, reply to requests, or spawn new agents. Activates on: peer messages, inbox, swarm, ccgram, broadcast, agent collaboration, ask another agent.
testing21
committing-code
Smart git commits with logical grouping. Use when user says "commit", "commit changes", "save changes", "create commit", "bundle commits", "git commit", or wants to commit their work.
development21
committing-code
Smart git commits with logical grouping. Use when user says "commit", "commit changes", "save changes", "create commit", "bundle commits", "git commit", or wants to commit their work.
development21
coding
Implementation process discipline for all languages — surface assumptions, define verifiable success criteria, and ground work in project domain docs. Use when implementing features, writing functions/classes/modules, or adding code. Complements language-specific skills and includes test-first guidance when implementation is explicitly TDD.
development21
refactoring-code
Batch refactoring via MorphLLM edit_file. Use for "refactor across files", "batch rename", "update pattern everywhere", large files (500+ lines), 5+ edits in same file, or applying an approved architecture-deepening refactor.
development21
learning-patterns
Extract learnings and generate Pi/project customizations (AGENTS.md, CONTEXT.md, ADRs, skills, agents, commands, hooks). Use when user says "learn", "extract learnings", "what did we learn", "save learnings", "adapt config", "capture domain language", or wants to improve Pi based on conversation patterns.
data-ai21
refactoring-code
Behavior-preserving refactoring in Pi using local search, edit, tests, and disciplined batching. Use when the user asks to refactor code without changing behavior.
development21
improve-codebase-architecture
Find deepening opportunities in a codebase, informed by the domain language in CONTEXT.md and the decisions in docs/adr/. Use when the user wants to improve architecture, find refactoring opportunities, consolidate tightly-coupled modules, or make a codebase more testable and AI-navigable. NOT for line-level cleanup (use reviewing-code) or batch edits (use refactoring-code).
development21
linting-instructions
Lint plugin agent/skill prompts against rules derived from Anthropic model cards (Opus 4.6, Sonnet 4.6). Use when authoring or reviewing skills and agents — "lint instructions", "audit prompts", "model card rules".
tools21
searching-code
Intelligent codebase search and zoom-out mapping via WarpGrep. Use when user asks "how does X work", "trace flow", "find all implementations", "understand codebase", "zoom out", "map this area", or needs cross-file exploration in large repos (1000+ files).
development21
reviewing-cc-config
Review Claude Code configuration for context efficiency, signal density, and anti-patterns. Use when user says "review config", "review setup", "check configuration", "review cc config", "context review", "config review", "review my setup", "review skills", "review agents", "review hooks", or wants feedback on their Claude Code configuration quality.
development21
looking-up-docs
Compatibility router for library documentation lookup. Use when user says "look up docs", "how to use", "API for", "syntax for", "examples of", "show me the docs", or needs API references, code examples, or framework-specific documentation. Routes to the context7-cli workflow.
tools21
documenting-code
Update project documentation based on recent changes. Use when user says "update docs", "document", "add documentation", "update readme", "write docs", or wants to improve documentation.
development21
ccgram-messaging
Inter-agent messaging via ccgram swarm. Use when communicating with other agents in the same tmux session — send messages, check inbox, discover peers, broadcast status, reply to requests, or spawn new agents. Activates on: peer messages, inbox, swarm, ccgram, broadcast, agent collaboration, ask another agent.
testing21
coding
Implementation process discipline for all languages — surface assumptions, define verifiable success criteria, and ground work in project domain docs. Use when implementing features, writing functions/classes/modules, or adding code. Complements language-specific skills and includes test-first guidance when implementation is explicitly TDD.
development21
using-git-worktrees
Creates isolated git worktrees for parallel development. Use when starting feature work needing isolation or working on multiple branches simultaneously. Not for simple branch switching or basic git operations.
development21
brainstorming-ideas
Brainstorm ideas and stress-test draft plans before coding. Use when user wants to brainstorm, explore approaches, design a feature/API/flow, challenge assumptions, or resolve terminology that blocks the design. NOT for pure decision-tree interviews on a single existing plan ("grill me"); use grill-me. NOT for breaking approved requirements into implementation tasks; use /spec:plan. NOT for general documentation updates; use documenting-code or learning-patterns.
development21
using-git-worktrees
Creates isolated git worktrees for parallel development. Use when starting feature work needing isolation or working on multiple branches simultaneously. Not for simple branch switching or basic git operations.
development21
sequential-thinking
Structured stepwise reasoning with explicit revisions and branches. Use when the user says "think step by step", "sequential thinking", "plan this out", "reason through this", "branch this idea", or when tackling a hard multi-step problem (architecture decisions, ambiguous bugs, multi-constraint tradeoffs, plans that may need revision). NOT for trivial lookups, single-tool fetches, or tasks the model can answer directly without planning.
tools21
researching-web
Web research via Pi web providers. Use for technical comparisons, recent facts, best practices, standards, pros and cons, or questions needing grounded web evidence.
development21
analyzing-usage
Analyze Claude Code usage, cost, efficiency, and burn rate using ccusage and termgraph. Use when user says "usage", "cost", "spending", "tokens", "analyze usage", "how much did I spend", "usage report", "budget", "burn rate", "efficiency", "cache hits", "ccusage", "ccw", "ccp".
development21
mem-history
Query project history and past decisions using local files and git history in Pi. Use when the user asks what changed, why a decision was made, or what the project already knows.
testing21
exploring-repos
Explore public GitHub repositories in Pi using GitHub CLI, local clones, and web tools. Use when the user asks how a public repo works, wants architecture orientation, or needs repo-level Q&A.
tools21
using-modern-cli
Prefer modern CLI tools for better performance: rg (ripgrep) instead of grep for text searching, fd instead of find for file discovery, bat instead of cat for viewing files, sd instead of sed for text replacement, eza instead of ls for directory listing, dust instead of du, procs instead of ps. Use when writing bash scripts, optimizing command chains, working with file searches, or replacing grep/find/cat in workflows. These tools are faster, respect .gitignore, and have better output formatting.
tools21
grill-me
Interview the user relentlessly about a plan or design until reaching shared understanding, resolving each branch of the decision tree. Use when user says "grill me", wants to stress-test a single plan, or asks to be challenged on a specific design. NOT for full ideation/feature design (use brainstorming-ideas) or thesis-vs-antithesis debates (use debating-ideas).
testing21
evolving-config
Audit AI coding-agent configuration against current features and local usage. Use when the user wants to improve Claude Code, Pi, Codex, Gemini, skill, hook, or agent configuration.
development21
debating-ideas
Dialectic thinking — spawn thesis and antithesis agents to stress-test ideas, then synthesize and verify against code. Use when user says "debate", "argue both sides", "devil's advocate", "stress test this idea", "pros and cons of approach", or wants rigorous evaluation of a design decision.
development21
context7-cli
Current library documentation via the ctx7 CLI. Use when the user mentions "ctx7" or "context7", asks for API docs, syntax, code examples, versioned library behavior, or needs docs lookup without provider-specific tools.
tools21
brainstorming-ideas
Brainstorm ideas and stress-test draft plans before coding. Use when brainstorming, exploring approaches, designing a feature/API/flow, grilling a plan, challenging assumptions, or resolving terminology that blocks the design. NOT for implementation task breakdown; use /spec:plan. NOT for general documentation updates; use documenting-code or learning-patterns.
development21
smart-explore
Token-efficient code navigation via AST parsing. Use when exploring file structure, cross-file symbol discovery, or targeted function extraction with smart_outline, smart_search, and smart_unfold. 10-20x fewer tokens than reading full files.
development21
sequential-thinking
Structured stepwise reasoning with explicit revisions and branches. Use when the user says "think step by step", "sequential thinking", "plan this out", "reason through this", "branch this idea", or when tackling a hard multi-step problem (architecture decisions, ambiguous bugs, multi-constraint tradeoffs, plans that may need revision). NOT for trivial lookups, single-tool fetches, or tasks the model can answer directly without planning.
tools21
reviewing-cc-config
Review Claude Code configuration for context efficiency, signal density, and anti-patterns. Use when user says "review config", "review setup", "check configuration", "review cc config", "context review", "config review", "review my setup", "review skills", "review agents", "review hooks", or wants feedback on their Claude Code configuration quality.
development21
mem-history
Query project history, past decisions, and known gotchas from claude-mem observations. Use when user asks "last session", "did we already", "what did we decide", "project history", "timeline", or "what happened with".
development21
debating-ideas
Dialectic thinking — spawn thesis and antithesis agents to stress-test ideas, then synthesize and verify against code. Use when user says "debate", "argue both sides", "devil's advocate", "stress test this idea", "pros and cons of approach", or wants rigorous evaluation of a design decision.
development21
researching-web
Web research via Perplexity AI. Use for technical comparisons (X vs Y), best practices, industry standards, recent developments. Triggers on "research", "compare", "vs", "best practice", "which is better", "pros and cons", "what's new in". NOT for API references or library docs (use looking-up-docs for syntax, examples, and framework documentation).
development21
looking-up-docs
Compatibility router for library documentation lookup. Use when user says "look up docs", "how to use", "API for", "syntax for", "examples of", "show me the docs", or needs API references, code examples, or framework-specific documentation. Routes to the context7-cli workflow.
tools21
learning-patterns
Extract learnings and generate project-specific customizations (CLAUDE.md, CONTEXT.md, ADRs, commands, skills, hooks). Use when user says "learn", "extract learnings", "what did we learn", "save learnings", "adapt config", "capture domain language", or wants to improve Claude Code based on conversation patterns.
development21
evolving-config
Audit Claude Code configuration against latest features and best practices. Use when user says "evolve", "self-improve", "audit config", "what's new in claude code", "upgrade configuration", "check for improvements", "are we up to date".
development21
brainstorming-ideas
Brainstorm ideas and stress-test draft plans before coding. Use when brainstorming, exploring approaches, designing a feature/API/flow, grilling a plan, challenging assumptions, or resolving terminology that blocks the design. NOT for implementation task breakdown; use /spec:plan. NOT for general documentation updates; use documenting-code or learning-patterns.
development21
analyzing-usage
Analyze Claude Code usage, cost, efficiency, and burn rate using ccusage and termgraph. Use when user says "usage", "cost", "spending", "tokens", "analyze usage", "how much did I spend", "usage report", "budget", "burn rate", "efficiency", "cache hits", "ccusage", "ccw", "ccp".
development21
reviewing-code
Sequential code review for security, quality, tests, and architecture. Use when reviewing code, checking changes, reviewing PRs, or looking for deep-module/refactoring opportunities.
development21
using-gemini
Consult Gemini for second opinions, brainstorming, or web search. Use when user says "ask gemini", "gemini search", "get gemini opinion", or wants a second AI perspective.
development21
playwright-skill
Internal Playwright helper library for Pi. Use when loaded by testing-e2e for dev-server detection, temporary browser scripts, screenshots, and helper utilities. Not directly invoked by users.
tools21
testing-e2e
Sequential E2E workflow for Pi. Use when running existing Playwright tests, generating browser checks, recording a visible session, or verifying a user flow end-to-end.
testing21
managing-infra
Infrastructure patterns for Kubernetes, Terraform, Helm, Kustomize, and GitHub Actions. Use when making K8s architectural decisions, choosing between Helm vs Kustomize, structuring Terraform modules, writing CI/CD workflows, or applying security best practices.
testing21
writing-go
Idiomatic Go 1.25+ development. Use when writing Go code, designing APIs, discussing Go patterns, or reviewing Go implementations. Emphasizes stdlib, concrete types, simple error handling, and minimal dependencies.
development21
writing-go
Idiomatic Go 1.25+ development. Use when writing Go code, designing APIs, discussing Go patterns, or reviewing Go implementations. Emphasizes stdlib, concrete types, simple error handling, and minimal dependencies.
development21
using-git-worktrees
Creates isolated git worktrees for parallel development. Use when starting feature work needing isolation or working on multiple branches simultaneously. Not for simple branch switching or basic git operations.
development21
spec
Spec-driven development for AI coding agents — captures requirements, builds epics with vertical-slice tasks, runs implementation one task at a time with user approval, and tracks evidence-based completion. Use when the user wants to start a structured project, capture requirements for a feature, plan an epic, work on the next task, mark a task done, check progress, or otherwise drive a project through REQ → EPIC → TASK artifacts under `.spec/`.
development21
mem-history
Query project history, past decisions, and known gotchas from claude-mem observations. Use when user asks "last session", "did we already", "what did we decide", "project history", "timeline", or "what happened with".
development21
linting-instructions
Lint plugin agent/skill prompts against rules derived from Anthropic model cards (Opus 4.6, Sonnet 4.6). Use when authoring or reviewing skills and agents — "lint instructions", "audit prompts", "model card rules".
tools21
spec
Spec-driven development for AI coding agents — captures requirements, builds epics with vertical-slice tasks, runs implementation one task at a time with user approval, and tracks evidence-based completion. Use when the user wants to start a structured project, capture requirements for a feature, plan an epic, work on the next task, mark a task done, check progress, or otherwise drive a project through REQ → EPIC → TASK artifacts under `.spec/`.
development21
using-gemini
Consult Gemini for second opinions, brainstorming, or web search. Use when user says "ask gemini", "gemini search", "get gemini opinion", or wants a second AI perspective.
development21
spec
Spec-driven development for AI coding agents — captures requirements, builds epics with vertical-slice tasks, runs implementation one task at a time with user approval, and tracks evidence-based completion. Use when the user wants to start a structured project, capture requirements for a feature, plan an epic, work on the next task, mark a task done, check progress, or otherwise drive a project through REQ → EPIC → TASK artifacts under `.spec/`.
development21
spec
Spec-driven development for AI coding agents — captures requirements, builds epics with vertical-slice tasks, runs implementation one task at a time with user approval, and tracks evidence-based completion. Use when the user wants to start a structured project, capture requirements for a feature, plan an epic, work on the next task, mark a task done, check progress, or otherwise drive a project through REQ → EPIC → TASK artifacts under `.spec/`.
development21
linting-instructions
Lint plugin agent/skill prompts against rules derived from Anthropic model cards (Opus 4.6, Sonnet 4.6). Use when authoring or reviewing skills and agents — "lint instructions", "audit prompts", "model card rules".
tools21
grill-me
Interview the user relentlessly about a plan or design until reaching shared understanding, resolving each branch of the decision tree. Use when user says "grill me", wants to stress-test a single plan, or asks to be challenged on a specific design. NOT for full ideation/feature design (use brainstorming-ideas) or thesis-vs-antithesis debates (use debating-ideas).
testing21
context7-cli
Current library documentation via the ctx7 CLI. Use when the user mentions "ctx7" or "context7", asks for API docs, syntax, code examples, versioned library behavior, or needs docs lookup without provider-specific tools.
tools21
analyzing-usage
Analyze AI coding agent usage, cost, efficiency, and burn rate — Claude Code (ccusage), Pi/pi-agent (ccusage-pi), or Codex CLI (ccusage-codex). Use when user says "usage", "cost", "spending", "tokens", "analyze usage", "how much did I spend", "usage report", "budget", "burn rate", "efficiency", "cache hits", "ccusage", "ccw", "ccp". NOT for general shell scripting, non-AI-agent cost analysis, or cloud infrastructure billing (use using-cloud-cli).
tools21
writing-typescript
Idiomatic TypeScript development. Use when writing TypeScript code, Node.js services, React apps, or discussing TS patterns. Emphasizes strict typing, composition, and modern tooling (bun/vite).
tools21
managing-infra
Infrastructure patterns for Kubernetes, Terraform, Helm, Kustomize, and GitHub Actions. Use when making K8s architectural decisions, choosing between Helm vs Kustomize, structuring Terraform modules, writing CI/CD workflows, or applying security best practices.
testing21
learning-patterns
Extract learnings and generate project-specific customizations (CLAUDE.md, CONTEXT.md, ADRs, commands, skills, hooks). Use when user says "learn", "extract learnings", "what did we learn", "save learnings", "adapt config", "capture domain language", or wants to improve Claude Code based on conversation patterns.
development20
using-gemini
Consult Gemini for second opinions, brainstorming, or web search. Use when user says "ask gemini", "gemini search", "get gemini opinion", or wants a second AI perspective.
development20
ccgram-messaging
Inter-agent messaging via ccgram swarm. Use when communicating with other agents in the same tmux session — send messages, check inbox, discover peers, broadcast status, reply to requests, or spawn new agents. Activates on: peer messages, inbox, swarm, ccgram, broadcast, agent collaboration, ask another agent.
testing20
coding
Implementation process discipline for all languages — surface assumptions, define verifiable success criteria, and ground work in project domain docs. Use when implementing features, writing functions/classes/modules, or adding code. Complements language-specific skills and includes test-first guidance when implementation is explicitly TDD.
development20
improve-codebase-architecture
Find deepening opportunities in a codebase, informed by the domain language in CONTEXT.md and the decisions in docs/adr/. Use when the user wants to improve architecture, find refactoring opportunities, consolidate tightly-coupled modules, or make a codebase more testable and AI-navigable. NOT for line-level cleanup (use reviewing-code) or batch edits (use refactoring-code).
development20
reviewing-code
Multi-agent code review for security, quality, architecture, and deep-module design. Use when user says "review code", "check code", "code review", "review my changes", "review this PR", "improve architecture", "find refactoring opportunities", "deep modules", or wants feedback on source code/design. NOT for Claude Code configuration review (use reviewing-cc-config for skills/agents/hooks/ CLAUDE.md review).
development20
managing-infra
Infrastructure patterns for Kubernetes, Terraform, Helm, Kustomize, and GitHub Actions. Use when making K8s architectural decisions, choosing between Helm vs Kustomize, structuring Terraform modules, writing CI/CD workflows, or applying security best practices.
testing20
writing-python
Idiomatic Python 3.12+ development. Use when writing Python code, CLI tools, scripts, or services. Emphasizes stdlib, type hints, uv/ruff/pyright toolchain, and minimal dependencies.
tools20
planning-review
Review implementation plans in docs/plans/ for correctness, scope, testing, and over-engineering. Use when user asks to review a plan before coding or validate plan quality.
testing20
planning-make
Create structured implementation plans in docs/plans/. Use when user asks to make a plan, implementation plan, rollout plan, or wants structured task breakdown before coding.
documentation20
planning-exec
Execute implementation plans task by task with strict verification and plan tracking. Use when user wants to run a plan, execute tasks from docs/plans/, or implement from an existing plan.
documentation20
context7-cli
Current library documentation via the ctx7 CLI. Use when the user mentions "ctx7" or "context7", asks for API docs, syntax, code examples, versioned library behavior, or needs docs lookup without provider-specific tools.
tools20
writing-typescript
Idiomatic TypeScript development. Use when writing TypeScript code, Node.js services, React apps, or discussing TS patterns. Emphasizes strict typing, composition, and modern tooling (bun/vite).
tools20
exploring-repos
Explore public GitHub repositories via DeepWiki AI-generated documentation. Use for understanding architecture, patterns, design decisions, and code organization of popular open-source projects. Use when user asks "how does X repo work", "explain architecture of Y", "what patterns does Z use", "explore repo", "deepwiki", or needs codebase-level understanding beyond API docs.
development20
debating-ideas
Dialectic thinking — spawn thesis and antithesis agents to stress-test ideas, then synthesize and verify against code. Use when user says "debate", "argue both sides", "devil's advocate", "stress test this idea", "pros and cons of approach", or wants rigorous evaluation of a design decision.
development20
writing-go
Idiomatic Go 1.25+ development. Use when writing Go code, designing APIs, discussing Go patterns, or reviewing Go implementations. Emphasizes stdlib, concrete types, simple error handling, and minimal dependencies.
development20
writing-web
Simple web development with HTML, CSS, JS, and HTMX. Auto-activates when working with .html, .css, or .htmx files, web templates, stylesheets, or vanilla JS scripts. NOT for React/Vue/Angular (use writing-typescript) or Node.js backends.
development20
playwright-skill
Internal Playwright automation library. Use when loaded by testing-e2e or playwright-tester for dev server detection, script execution, and browser automation primitives. Not directly invoked by users.
tools20
testing-e2e
E2E testing with Playwright — the primary user-facing skill for writing, running, and generating browser tests. Use when user asks to "write e2e tests", "test this page", "run playwright tests", "generate browser tests", "check accessibility", or "visual regression". Supports TypeScript tests and Go/HTMX web applications.
development20
searching-code
Intelligent codebase search and zoom-out mapping via WarpGrep. Use when user asks "how does X work", "trace flow", "find all implementations", "understand codebase", "zoom out", "map this area", or needs cross-file exploration in large repos (1000+ files).
development20
refactoring-code
Batch refactoring via MorphLLM edit_file. Use for "refactor across files", "batch rename", "update pattern everywhere", large files (500+ lines), 5+ edits in same file, or applying an approved architecture-deepening refactor.
development20
documenting-code
Update project documentation based on recent changes. Use when user says "update docs", "document", "add documentation", "update readme", "write docs", or wants to improve documentation.
development20
sequential-thinking
Structured stepwise reasoning with explicit revisions and branches. Use when the user says "think step by step", "sequential thinking", "plan this out", "reason through this", "branch this idea", or when tackling a hard multi-step problem (architecture decisions, ambiguous bugs, multi-constraint tradeoffs, plans that may need revision). NOT for trivial lookups, single-tool fetches, or tasks the model can answer directly without planning.
tools20
evolving-config
Audit Claude Code configuration against latest features and best practices. Use when user says "evolve", "self-improve", "audit config", "what's new in claude code", "upgrade configuration", "check for improvements", "are we up to date".
development20
mem-history
Query project history, past decisions, and known gotchas from claude-mem observations. Use when user asks "last session", "did we already", "what did we decide", "project history", "timeline", or "what happened with".
development20
smart-explore
Token-efficient code navigation via AST parsing. Use when exploring file structure, cross-file symbol discovery, or targeted function extraction with smart_outline, smart_search, and smart_unfold. 10-20x fewer tokens than reading full files.
development20
using-modern-cli
Prefer modern CLI tools for better performance: rg (ripgrep) instead of grep for text searching, fd instead of find for file discovery, bat instead of cat for viewing files, sd instead of sed for text replacement, eza instead of ls for directory listing, dust instead of du, procs instead of ps. Use when writing bash scripts, optimizing command chains, working with file searches, or replacing grep/find/cat in workflows. These tools are faster, respect .gitignore, and have better output formatting.
tools20
analyzing-usage
Analyze Claude Code usage, cost, efficiency, and burn rate using ccusage and termgraph. Use when user says "usage", "cost", "spending", "tokens", "analyze usage", "how much did I spend", "usage report", "budget", "burn rate", "efficiency", "cache hits", "ccusage", "ccw", "ccp".
development20
reviewing-cc-config
Review Claude Code configuration for context efficiency, signal density, and anti-patterns. Use when user says "review config", "review setup", "check configuration", "review cc config", "context review", "config review", "review my setup", "review skills", "review agents", "review hooks", or wants feedback on their Claude Code configuration quality.
development20
researching-web
Web research via Perplexity AI. Use for technical comparisons (X vs Y), best practices, industry standards, recent developments. Triggers on "research", "compare", "vs", "best practice", "which is better", "pros and cons", "what's new in". NOT for API references or library docs (use looking-up-docs for syntax, examples, and framework documentation).
development20
committing-code
Smart git commits with logical grouping. Use when user says "commit", "commit changes", "save changes", "create commit", "bundle commits", "git commit", or wants to commit their work.
development20
improving-tests
Improve test design and coverage, including TDD/red-green-refactor guidance. Use when user says "improve tests", "refactor tests", "test coverage", "combine tests", "table-driven", "parametrize", "test.each", "test-first", "TDD", "red-green-refactor", or wants to remove test waste.
development20
looking-up-docs
Compatibility router for library documentation lookup. Use when user says "look up docs", "how to use", "API for", "syntax for", "examples of", "show me the docs", or needs API references, code examples, or framework-specific documentation. Routes to the context7-cli workflow.
tools20
deploying-infra
Validate and deploy Kubernetes, Terraform, Helm, Kustomize, GitHub Actions, and Docker configs. Use when user says "deploy", "deploy to staging", "apply changes", "terraform apply", "helm upgrade", "kubectl apply", "rollout", "deploy check", "validate deployment", "validate infrastructure", or wants to verify or apply infrastructure changes.
testing19
fixing-code
Fix code problems with disciplined diagnosis — run checks, build a repro for bugs, rank falsifiable hypotheses, fix one issue at a time, and verify until clean. Use when user says "fix", "debug", "diagnose", "broken", "failing", "throws", "performance regression", "make tests pass", or wants lint/test/build failures resolved.
development19
grill-me
Interview the user relentlessly about a plan or design until reaching shared understanding, resolving each branch of the decision tree. Use when user says "grill me", wants to stress-test a single plan, or asks to be challenged on a specific design. NOT for full ideation/feature design (use brainstorming-ideas) or thesis-vs-antithesis debates (use debating-ideas).
testing19