Skill Inventory & Patterns
Personal inventory of existing skills and cross-cutting patterns.
Local Skills (research-lab/.claude/skills/)
| Skill | Category | Sophistication | Key Pattern |
|---|---|---|---|
proposal | Workflow | High | Lifecycle state machine (IDEATE->SYNTHESIZE) |
research | Workflow | Very High | Multi-AI investigation + prompt enrichment |
learn | Document | High | Progressive complexity (concepts->experiments) |
design-arena | Workflow | Extreme | Multi-model tmux orchestration + Caddy proxy |
issue | Workflow | High | Dual-mode (investigation vs. discovery) |
pr-review | Workflow | Extreme | 3 parallel review subagents + tech-debt lifecycle |
explore | Document | High | Profile-aware code-grounded analysis |
swarm | Workflow | Extreme | Two-pass hybrid (tmux agents + Task subagents) |
setup-repo | Workflow | Medium | SSH-first bootstrapping |
profile | Workflow | High | GitHub-aware temporal staleness detection |
Global Skills (~/.claude/skills/)
| Skill | Category | Sophistication | Key Pattern |
|---|---|---|---|
agi | Workflow | Very High | Autonomous tmux agents with session persistence |
ai-review | Workflow | High | Confidence x severity matrix, language checklists |
breakdown-plan | Workflow | High | Phase decomposition with entrypoint prompt |
committee | Workflow | High | Multi-model deliberation with synthesis |
implement-plan | Workflow | High | Sequential subagent execution per phase |
initial-plan | Workflow | High | Proposal-to-plan with auto-review pass |
orchestrate-plan | Workflow | Extreme | End-to-end autonomous pipeline (5 chained skills) |
review-breakdown | Workflow | Very High | Dual-lens parallel review + direct fix application |
redate-commits | Workflow | Very High | GitHub activity analysis + proportional gap distribution |
tmux | Workflow | Medium | Foundation patterns for agent spawning |
Cross-Cutting Patterns
Auto-bootstrap: Check if infrastructure exists, create if not (setup-repo, issue, explore, swarm)
Code-grounded analysis: Reference actual file paths and line numbers, not summaries (issue, explore, pr-review)
Multi-AI orchestration: Run multiple models in parallel via tmux (research, design-arena, swarm, agi, committee)
Sentinel files: Monitor completion via marker files instead of parsing output (design-arena, swarm)
Pipe-prompt delivery: cat /tmp/prompt.md | claude -p to avoid shell escaping (swarm)
Archival strategy: Current round always has clean names; previous rounds renamed to -prev-{N} (agi, committee)
Dual-lens review: Two complementary review perspectives in parallel, then consolidation (pr-review, review-breakdown)
Progressive disclosure in practice: All 20 skills keep SKILL.md self-contained. None use scripts/ or references/ directories. All logic is embedded in the markdown.
Notable Techniques
- tmux send-keys pattern: Always send text and Enter as separate commands with 0.3s sleep between
unset CLAUDECODE: Required before spawning Claude in tmux (prevents nested-instance error)- Pane ID capture: Use plain variables, not arrays (bash/zsh indexing differs)
- Destructive command audit: orchestrate-plan scans for dangerous commands before running
- Temp file tracking: Log temp files, add to .gitignore, clean up at end