How coleam00/skills works
Cole’s AI Skills is a collection of editable agent procedures, with executable Claude Code hooks as a complementary runtime layer. The principal workflow is the README’s ticket loop: prime context, plan, implement, validate, review, commit, and open a PR. Other documented capabilities cover product planning, issue diagnosis, parallel worktrees, skill/rule creation, autonomous factory setup, and agent tools. Skill bodies and hooks/README.md were not sampled, so those subsystems are represented at directory level; hook behavior and its effects are included only where source excerpts show them.
Main entry points: piv-plan-implementation Plan implementation · piv-implement Implement · piv-validate Validate · piv-review-changes Review changes
Components 34
Ticket workflow
- Prime context
.claude/skills/prime-codebase - Plan implementation
.claude/skills/piv-plan-implementation - Implement
.claude/skills/piv-implement - Validate
.claude/skills/piv-validate - Review changes
.claude/skills/piv-review-changes - Commit
.claude/skills/piv-commit - Create PR
.claude/skills/piv-create-pr - Run full loop
.claude/skills/piv-run-full-loop
Planning and context
- Create PRD
.claude/skills/plan-create-prd - Plan architecture
.claude/skills/plan-architecture - Create stories
.claude/skills/plan-create-stories - Slice epic
.claude/skills/piv-slice-epic - Investigate issues
.claude/skills/piv-investigate-issue - Implement issue
.claude/skills/piv-implement-issue - Parallel worktrees
.claude/skills/worktree-create - Merge worktrees
.claude/skills/worktree-merge
AI layer building
- Rules and drift
.claude/skills/rules-create-global - Ablate AI layer
.claude/skills/ablate-ai-layer - Author skills
.claude/skills/skills-create - Create hooks
.claude/skills/hooks-create - Process review
.claude/skills/system-evolution-review
Tools and autonomy
- Dark factory
.claude/skills/build-dark-factory - Browser automation
.claude/skills/agent-browser - AST search
.claude/skills/ast-grep - Desktop control
.claude/skills/drive-screen - Setup tutor
.claude/skills/setup-ai-tutor
Runtime hooks
- Session context
hooks/session_start_context.py - Dependency guard
hooks/pre_tool_use_dependencies.py - Action audit log
hooks/post_tool_use_log.py - Test stop gate
hooks/stop_tests_must_pass.py - Stop notification
hooks/stop_notify.py
Other components
- Developer
- Coding agent
- Git repository
Connections 36
- Developer → Coding agent: selects skill
README.md - Prime context → Plan implementation: then plan
README.md - Plan implementation → Implement: then implement
README.md - Implement → Validate: then validate
README.md - Validate → Review changes: then review
README.md - Review changes → Commit: then commit
README.md - Commit → Create PR: then open PR
README.md - Coding agent → Run full loop: chains workflow
README.md - Developer → Create PRD: requests planning
README.md - Developer → Plan architecture: requests planning
README.md - Developer → Create stories: requests backlog
README.md - Developer → Slice epic: requests slicing
README.md - Developer → Investigate issues: reports issue
README.md - Investigate issues → Implement issue: informs fix
README.md - Developer → Parallel worktrees: requests parallel work
README.md - Parallel worktrees → Merge worktrees: integrates branches
README.md - Developer → Rules and drift: builds AI layer
README.md - Developer → Ablate AI layer: evaluates rules
README.md - Developer → Author skills: authors procedures
README.md - Developer → Create hooks: authors hooks
README.md - Developer → Process review: reviews process
README.md - Developer → Dark factory: builds autonomy
README.md - Coding agent → Browser automation: uses browser
README.md - Coding agent → AST search: searches syntax
README.md - Coding agent → Desktop control: controls desktop
README.md - Developer → Setup tutor: sets up sample
README.md - Git repository → Session context: provides git state
hooks/session_start_context.py - Session context → Coding agent: injects context
hooks/session_start_context.py - Coding agent → Dependency guard: reads or edits
hooks/pre_tool_use_dependencies.py - Dependency guard → Coding agent: blocks or injects
hooks/pre_tool_use_dependencies.py - Coding agent → Action audit log: records tool event
hooks/post_tool_use_log.py - Action audit log → Git repository: writes project log
hooks/post_tool_use_log.py - Coding agent → Test stop gate: attempts to stop
hooks/stop_tests_must_pass.py - Test stop gate → Coding agent: returns test failure
hooks/stop_tests_must_pass.py - Coding agent → Stop notification: finishes turn
hooks/stop_notify.py - Stop notification → Developer: sends notification
hooks/stop_notify.py