
Create, update, or optimize skills and SKILL.md content.
Automatically use for durable knowledge, configured project-context lookup, and schema-backed artifact layouts.
Manage Git preflight, branch and worktree creation, and completed-work cleanup. Use when explicitly invoked.
Route any request containing literal `trigger:` before ordinary skill dispatch. Read at thread start and whenever `trigger` appears.
Create, rename, audit, or close a tracked Codex task whose turns and status are persisted in the local thread ledger. Only use when directly invoked.
Finalize completed PR or local checkout work. Use when explicitly invoked
Create or audit Agent Project Directory docs and status structure.
List, inspect, validate, or materialize bundled file schemas.
Use only when explicitly invoked to manage ~/integ/[project] harness repos.
Create structured specs and code exploration docs.
Run explicitly requested swarm workflows for feature delivery.
Run live OpenClaw integration proof against a named claw gateway profile with showboat-v2.
Watch a pull request and CI until green or needing fixes.
Read a Markdown task file, parse frontmatter, split work by level-2 headings outside code fences, and execute each section with subagents while keeping level-3 headings as ordered follow-on instructions for the same subagent.
Improve skills from observed agent friction in sessions, PRs, or audits.
Turn response annotations and review comments into a question-and-answer FAQ.
Create, inspect, remove, or transactionally clean landed Git worktrees.
Record, sync, and query local agent activity ledger entries.
Create a self-contained interactive HTML explanation of a code diff, branch, commit, or pull request.
Fork or create a Codex thread from a prompt, with optional model and reasoning overrides. Fork by default; create without parent history when the user explicitly asks for a new, fresh, or clean thread.
Refactor an existing OpenClaw docs page with source-audited preservation, restructuring, and verification.
Top-level agents only: send exactly one final-state desktop notification before the final report.
Review code, docs, specs, architecture, UX, or design docs.
Load reusable reference docs for coding-related agent work. Always load in context at beginning of session.
Install or document local command-line tools end to end.
Create schema-backed Showboat integration proofs with scenario summaries, raw artifacts, and replayable verification.
Search prior Codex conversations and return up to 3 relevant thread matches with deep links. Use when explicitly invoked as `$convosearch [query] | [filters]`.
Land completed work and update AGD status docs.
Build, publish, and verify OpenClaw Docker/GHCR images. Use when the user asks to build an OpenClaw image, publish or republish GHCR tags, bundle official plugins such as @openclaw/codex, @openclaw/slack, or @openclaw/diagnostics-otel, make or verify multi-arch linux/amd64 and linux/arm64 images, check GHCR visibility, secret-scan an image, enforce Codex image contract gates, or prove a published OpenClaw Docker image works.
Audit an OpenClaw maturity-scorecard surface into an evidence-backed component score report. Use when given a surface from an OpenClaw maturity-scorecard.md and asked to score coverage, quality, readiness, or generate a detailed surface report plus per-component subreports.
Generate project icons, logos, or inline brand marks.
Create or improve Hacker News submission titles.
Link the current Codex session to a durable Markdown task note.
Use 1Password CLI safely for agent credential workflows.
Create, edit, extract, or analyze PDF documents.
Summarize Codex activity from the agent ledger.
Create executable Showboat demos that prove behavior.
Send explicit Slack notifications through `slack-post`.
Upload existing images to ImageKit and return CDN URLs.
Create, edit, analyze, or visualize spreadsheet files.
Add an Apache 2.0 license and update package metadata.
Proofread publish-ready drafts for clarity, correctness, and links.
Sweep recent OpenClaw GitHub issues, filter for real unstarted bugs, and write a report to $mem claw/main reports.
Implement, fix, refactor, or otherwise modify source code.
Run in fast-mode with only explicitly allowed skills.
Audit documentation rewrites with JSON-first, block-scoped, line-level coverage mappings. Use for moved-section checklists, preservation proofs, migration maps, or line-by-line coverage reviews of Markdown/MDX docs.
Turn an existing concrete spec into a reusable generic spec template. Use when asked to create a generic spec, template spec, reusable implementation template, or generalized version of a spec from a specific implementation such as one plugin, channel, integration, feature, or PR.
Write or review high-quality OpenClaw developer documentation.
Trace how something works with an investigator subagent and a skeptical reviewer subagent.
Create or revise engineering diagrams in ASCII or Mermaid.
Write or review Stripe-style developer documentation.
Audit a repo for code quality issues using focused heuristics.
Determine whether test failures come from the current branch.
Create tracked tasks or issues with session context.
Manage Codex self-configuration and session analysis; not skill authoring (use $sc).
Scan a repo for sensitive data before making it public.
Drive a development task through planning, execution, and verification.
Create, edit, review, or extract Word docx documents.
Manage Linear issues and projects through `linear-cli`.
Simulate implementing a spec against the real source code, then grade the spec as correct, comprehensive, and simple. Use only when explicitly invoked as $spec-simulate or when the user asks to simulate implementation from a spec.
Write repository-appropriate git commit messages by first checking local history, then matching that repo's conventions. Use when asked to draft a commit message, explain how a repo formats commits, or adapt commit style across different repositories.
Write sudocode from real code for docs and specs.
Generate incremental Slack digests for channels, topics, and categories.
Add or update CLI maturity ratings in repo docs.
Optimize skill instructions, resources, and usability.
Coordinate explicitly requested subagent work as a manager.
Read a Markdown task file, parse frontmatter, split work by level-2 headings outside code fences, and execute each section with subagents while keeping level-3 headings as ordered follow-on instructions for the same subagent.
Reproduce and fix OpenClaw issues with Showboat proof, managed implementation, and repo-local integration gateways.
Write or review technical documentation.
Use only when explicitly invoked to manage ~/integ/[project] harness repos.
Automate browser navigation, interaction, screenshots, and extraction.
Judge agent work as approve, reject, or escalate.
Use when creating TypeScript projects or add TypeScript pre-commit checks.
Draft, validate, or explain Conventional Commit messages.
Fill TODO, placeholder, or missing Markdown links.
Find, map, or resume Codex CLI sessions.
Run a development task from intake through implementation.
Refresh a repo by syncing local master from origin.
Load local dotenvx credential sets for agent workflows.