1---2name: commands-43description: Quick reference of all custom skills, agents, and key workflows4---5
6# /commands — BrickTrack Skill Reference
7
8Print this cheat sheet for the user. Do NOT execute any skills — just display the reference.
9
10## Development Skills
11
12| Command | Purpose |
13|---------|---------|
14| `/tdd-workflow` | Full red-green-refactor cycle with test-first development |
15| `/new-router` | Scaffold tRPC router + repository + interface + tests |
16| `/new-component` | Create React component following project conventions |
17| `/new-e2e-test` | Create Playwright E2E test with auth fixture + cleanup |
18| `/env-spinup` | Docker + DB + schema push + test validation |
19
20## Workflow Skills
21
22| Command | Purpose |
23|---------|---------|
24| `/task-complete` | bd close, create PR, merge, worktree cleanup |
25| `/document-continue` | Save progress before /clear for context handoff |
26| `/phase-transition` | Move between RESEARCH → PLAN → IMPLEMENT → REVIEW → VERIFY |
27| `/plan-to-epic` | Parse markdown plan into Beads tasks with auto-dependencies |
28| `/ralph-execute` | Autonomous epic execution with subagents + circuit breaker |
29| `/ralph-resume` | Resume a PAUSED Ralph session — retry, skip, or abort |
30| `/wrap-up` | End-of-session routine: commit, memory review, self-improvement |
31| `/commands` | This cheat sheet |
32
33## Quality & Review Skills
34
35| Command | Purpose |
36|---------|---------|
37| `/promote` | Verify staging → PR develop→main → merge → verify prod |
38| `/audit-prod` | Smoke test production pages, console errors, network failures |
39| `/eval-report` | Session metrics dashboard (pass@k, error rate, trends) |
40| `/agent-shield` | Security scan of .claude/ configuration files |
41
42## Learning & Meta Skills
43
44| Command | Purpose |
45|---------|---------|
46| `/evolve` | Cluster instincts, promote high-confidence to skills, prune stale |
47| `/skill-create` | Auto-generate skills from git history + instincts |
48| `/instinct-export` | Export instincts as portable JSONL for cross-project sharing |
49| `/instinct-import` | Import instincts from another project's export |
50| `/model-routing` | Display subagent model selection strategy + cost reference |
51
52## Search & Navigation Skills
53
54| Command | Purpose |
55|---------|---------|
56| `/conversation-search` | Search past session history by keyword with date filters |
57| `/conversation-fork` | Diverge into alternative approach via stash or worktree |
58| `/iterative-retrieve` | 4-phase context retrieval protocol for complex questions |
59
60## Custom Agents (auto-dispatched via Task tool)
61
62| Agent | Trigger |
63|-------|---------|
64| `code-reviewer` | After implementing features, before PRs |
65| `test-writer` | When tests needed for new/existing code |
66| `security-reviewer` | After API endpoints, auth changes, payments |
67| `database-reviewer` | After schema changes, new queries, migrations |
68| `build-error-resolver` | When tsc/build/tests fail |
69
70## Product & Strategy Skills
71
72| Command | Purpose |
73|---------|---------|
74| `/value-check` | Evaluate whether a feature or product idea delivers discoverable user value |
75
76## New Skills
77
78| Command | Purpose |
79|---------|---------|
80| `/debt-scan` | Scan for anti-patterns with counts, locations, and trend tracking |
81| `/session-search` | Search past session JSONL files by keyword |
82| `/review-claudemd` | Audit CLAUDE.md files for dead refs, deprecated APIs, token bloat |
83| `/gha` | Trigger/monitor GitHub Actions workflows via gh CLI |
84| `/rules-doctor` | Scan .claude/rules/ for broken paths and coverage gaps |
85| `/web-fallback` | Resilient web fetch with WebSearch fallback chain |
86
87## Built-in Claude Code Shortcuts
88
89| Shortcut | What it does |
90|----------|-------------|
91| `ESC ESC` | **Rewind conversation** — undo last exchange. Choose to keep code changes or revert both. Use after fixing side bugs to reclaim context. |
92| `/chrome` | Native browser integration for visual verification. Add accessibility tree guidance to CLAUDE.md (use `read_page` + `ref`, not coordinates). |
93| `/compact` | Summarize conversation to free context. Prefer `/document-continue` first. |
94| `/clear` | Start fresh conversation. Use between unrelated tasks or phase transitions. |
95| `Shift+Tab` | Toggle plan mode for architecture-first thinking. |
96| `Tab` | Accept autocomplete suggestion. |
97| `/plugin` | Open plugin manager — browse, install, manage, update plugins. |
98| `/plugin install <name>` | Install a plugin (e.g., `repomix`, `plannotator`, `playground`). |
99| `/memory` | View and manage auto-memory entries. Periodically nudge Claude to update these. |
100
101## Verification Techniques
102
103| Technique | When to use |
104|-----------|-------------|
105| **Playwright MCP** | E2E test verification, browser automation |
106| **`/chrome` browser** | Visual verification of UI changes (use `read_page` for element refs) |
107| **tmux session** | Verify interactive CLIs — `tmux new -d -s test`, send commands, capture output |
108| **Draft PR review** | `gh pr create --draft` — review all changes before marking ready |
109| **git bisect** | Find exact commit that broke something — Claude can automate with a test script |
110
111## Shell Aliases (after `source .claude/setup-aliases.sh`)
112
113| Alias | Expands To |
114|-------|------------|
115| `c` | `claude` |
116| `cc` | `claude --continue` |
117| `cr` | `claude --resume` |
118| `claude-dev` | Claude with development context |
119| `claude-review` | Claude with review context |
120| `claude-research` | Claude with research context |
121| `wt` | `cd ~/worktrees/property-tracker` |
122| `bdr` | `bd ready` |
majiayu000