samwang0723
26 verified skills26 total stars
arch-review
Run a multi-agent architecture review producing codebase map, mermaid diagrams, STRIDE threat model, and operational risk assessment. Use when the user says 'arch review', 'architecture audit', 'review the architecture', 'system design review', or needs a comprehensive assessment of codebase structure, dependencies, and security posture.
development1
doc-update
Review and update project documentation (README.md, CLAUDE.md, docs/) to reflect the current codebase. Use when the user asks to "update docs", "sync documentation", "refresh README", "update CLAUDE.md", or after significant code changes that may have made documentation stale. Also trigger when the user mentions "docs are outdated" or "documentation drift".
development1
effective-rails
Ruby on Rails coding best practices for idiomatic, efficient, and maintainable code. Use when writing Ruby on Rails code, reviewing code, or learning Rails patterns.
development1
effective-rust
Expert Rust developer specializing in ownership semantics, zero-cost abstractions, and idiomatic patterns. This skill should be used PROACTIVELY when working on any Rust code - implementing features, debugging borrow checker issues, optimizing performance, or reviewing code quality. Use unless a more specific subagent role applies.
development1
knowledge
Search, browse, and update the Obsidian knowledge vault using the Obsidian CLI. Use when asked about SOPs, interview questions, engineering references, project docs, commands, or anything that might be in personal notes.
tools1
lead-cleanup
Remove git worktrees and branches for a completed task. Use after merging the integrate branch. Confirms before destructive action.
tools1
learn
Extract reusable patterns from the current session and save them as skill files. Use when the user says 'learn from this', 'extract patterns', 'save this as a skill', 'what did we learn', or after solving a non-trivial debugging or integration problem worth capturing for future sessions.
development1
market-research
Conduct market research, competitive analysis, investor due diligence, and industry intelligence with source attribution and decision-oriented summaries. Use when the user wants market sizing, competitor comparisons, fund research, technology scans, or research that informs business decisions.
testing1
review-pr
PR review pipeline. Dev review, peer review, QA, security, architecture gate. Use when told to "review PR", "review changes", "check this PR".
development1
scope
Project scoping. Requirements, estimates, design, early security flags. Use when asked to "scope", "estimate", "how long", "plan this".
testing1
secure-config
Security audit for Claude Code configuration. Scans agents, skills, rules, hooks, settings, and MCP servers for malware, data exfiltration, prompt injection, or system compromise. Use when the user says 'audit config', 'check for malware', 'security scan', 'is my setup safe', 'scan hooks', 'check skills for backdoors', or installs new skills/agents from untrusted sources.
tools1
strategy
Strategic technical decision. Build-vs-buy, technology choices. Use when asked "should we", "compare", "build vs buy", "evaluate".
development1
team-start
Spawn an agent team with multiple parallel Claude instances in tmux panes. Each teammate is a separate session with its own context window. Creates git worktrees and markdown state tracking alongside JSON tasks. Supports: --agents (comma-separated), --members N, --devs N. Use when told to "team up", "parallel team", "spawn a team", or "standby".
development1
write-prd
Write a Product Requirements Document (PRD) using the standard TMAB template with stakeholders, user stories (Given-When-Then), success metrics, and A/B testing plans. Use when the user asks to write a PRD, create product requirements, document a feature spec, or plan a new product feature.
testing1
effective-go
Idiomatic Go patterns for production services: error handling, HTTP/gRPC handlers, concurrency, testing, database, and observability. Use PROACTIVELY when writing, reviewing, or refactoring Go code.
development1
api-design
REST API design patterns including resource naming, status codes, pagination, filtering, error responses, versioning, and rate limiting for production APIs.
development1
investigate
Incident investigation. Logs, changes, root cause, blast radius. Use when told to "investigate", "debug", "what happened", "why broken".
development1
effective-typescript
Idiomatic TypeScript patterns for production applications: strict typing, error handling, async patterns, React/Next.js, Node.js services, testing, and performance. Use PROACTIVELY when writing, reviewing, or refactoring TypeScript code.
development1
pr-comments
Review and address PR review comments. Fetches comments from a GitHub PR, presents them in a structured table, and asks the user which to address before making any code changes. Use when the user says "check PR comments", "address review feedback", "fix PR comments", or "pr-comments".
development1
strategic-compact
Suggest manual /compact at logical workflow boundaries instead of relying on arbitrary auto-compaction. Use when the session approaches context limits, after completing a major milestone, when switching between unrelated tasks, or when responses become less coherent due to context pressure. Pairs with the suggest-compact.sh hook for automated threshold detection.
data-ai1
team-status
Check progress of an active agent team. Shows member status, completed tasks, pending work, and any messages. Use when asked "team status", "how's the team", "check team progress", "team update".
testing1
write-tech-spec
Generate a technical specification document using the DDD template in template.md. Use when the user says 'write tech spec', 'create tech spec', 'technical specification', or needs a structured design document for a new feature or major change covering architecture, domain models, APIs, data design, security, and operations.
development1
lead-summary
Quick progress summary of all tasks. Pipeline status, blockers, next actions. Use when asked for "status", "progress", "summary", "blockers", "what's happening".
devops1
quick-scan
Fast codebase health check. Structure, stack, dependencies, test status. Use when asked to "scan", "health check", "quick look".
development1
lead-start
Start a new engineering task or resume an existing one. Main entry point. Creates task folder, worktrees, delegates through team-lead. Supports --devs N (1-5). Use when told to "start", "build", "implement", "work on", "resume", or a task slug is referenced.
development1
team-stop
Clean up an agent team. Removes team resources, optionally cleans worktrees and branches. Use after team work is complete and merged. Use when told to "stop team", "cleanup team", "disband team".
data-ai1