# SKILLS — Master Index

One-page map of every skill across the three lanes. Trigger phrase → skill → when to use.

**Append-only.** New skills added at the bottom of their lane section. Don't reshuffle — agents bookmark line ranges.

**Last updated:** 2026-05-06 (initial draft)

---

## Conventions

- **Trigger** column = exact phrase or pattern that activates the skill. Multiple triggers separated by `/`.
- **When to use** = the single decisive condition. If unsure, read the skill's `SKILL.md`.
- **Path** = where `SKILL.md` lives. Vault-scoped skills live under their lane; user-scoped (cross-project) live in `~/.claude/skills/`.
- **Status:** `live` (working), `draft` (skeleton only), `v2` (planned, not built).
- **Calls:** programmatic skills invoked by other skills are marked `(programmatic)` — not user-facing.

---

## Lane: 10_RateRight (growth, hunt, marketplace)

User-level skills, invoked from any working dir. Live at `~/.claude/skills/`.

| Skill | Trigger | When to use | Status |
|---|---|---|---|
| `hunt` | `/hunt` | Kick off marketplace hunter scrape cycle (Facebook Marketplace, Gumtree, etc. → leads). | live |
| `matchmaker` | `/matchmaker` | Run match engine + manage outreach queue (approve leads, check stats, toggle automation). | live |
| `post` | `/post` | Compose + execute Facebook group post via headed browser; track history. | live |
| `pulse` | `/pulse` | One-command health check of marketplace hunter system. Compact status dashboard. | live |

**Vault-side companion docs:** `10_RateRight/_Brain/CLAUDE.md`, `10_RateRight/Skills/_README.md` (catalog), `10_RateRight/Skills/_Log.md` (per-skill version history), `10_RateRight/Skills/<name>/SKILL.md` (vault mirror of user-level), `10_RateRight/Skills/<name>/fixtures/golden-output.md` (regression fixture stub per skill).

**Sync convention:** user-level (`~/.claude/skills/<name>.md`) is canonical for invocation. Vault mirrors are for fixture/discoverability. Edit user-level first, then mirror.

---

## Lane: 11_OpsMan_LFCS (Hermes, OpsMan, LFCS bid pipeline)

Vault-scoped skills under `HQ-Vault/11_OpsMan_LFCS/Operations/skills/`. Bid-intake pipeline is the dominant chain — see ordering note below.

### Bid-intake pipeline (run in order)

| # | Skill | Trigger | When to use | Status |
|---|---|---|---|---|
| 0 | `lfcs-job-scaffold` | "New bid for [job#] [client] [site]" / "Awarded — set up the project folder" | Brand-new job arrives. Creates 11-section folder + seeds `00-brief.md` + `README.md`. | live |
| 1 | `lfcs-drawing-extract-text` | "Extract drawings for [job]" / "Build the drawing index" / "I dropped the drawings" | Pass 1 — text-only PDF extraction → `Drawing-Index.md`. Cheap, fast, runs on every drawing PDF in `03a`. | live |
| 2 | `lfcs-drawing-extract-vision` | "Vision-pass [job]" / "Upgrade Drawing-Index.md to High confidence" | Pass 2 — page-by-page vision read for drawn-only content (dimensions, hatching, callouts). Mandatory before pricing finalisation. | live |
| 3 | `lfcs-boq-parse` | "Parse the BOQ for [job]" / "I dropped the BOQ" / "Build quantities table" | Bill of Quantities xlsx → normalised JSON + at-a-glance MD table. Feeds bid-prep + incl-excl. | live |
| 4 | `lfcs-spec-summarise` | "Summarise spec [path or code]" / "Build spec summary for R11" | Mirror a spec PDF (TfNSW R-series, Sydney Water, AS/NZS) to a structured MD summary so downstream skills don't re-read the PDF every run. | live |
| 5 | `lfcs-bid-prep-doc` | "Build the bid prep doc for [job]" / "Generate 00-Bid-Prep" | Once drawings + BOQ extracted, generate canonical `00-Bid-Prep.md`. | live |
| 6 | `lfcs-precall-questions` | "Build the QS call questions for [job]" / "QS call coming up — what do I need to ask?" | After bid-prep, generate `Pre-Call-Questions-{date}.md` agenda from gaps/risks. | live |
| 7 | `lfcs-incl-excl-draft` | "Draft inclusions and exclusions for [job]" / "Build the incl-excl doc" | After pre-call questions, draft `Inclusions-Exclusions.md`. Refined post-Tuesday QS call. | live |
| 4.5 | `lfcs-pre-grill` | "Build pre-pricing grill for [job]" / "Run Phase 0 for [job]" / "Pre-grill [job]" | After BoQ parsed, before bid-prep-doc. Codifies Phase 0 mandatory grill from pricing-checklist.md. Aborts pricing on TBC answers. | live |

### Orchestrator (one-shot full intake)

| Skill | Trigger | When to use | Status |
|---|---|---|---|
| `lfcs-project-init` | "New job arrived — set me up" / "Init project [job#]" | **Single-skill orchestrator.** Runs scaffold → drawings → BOQ → bid-prep → pre-call → incl-excl in order with abort-on-gap. Use this instead of stepping through pipeline manually. | live |

### Programmatic / cross-cutting

| Skill | Trigger | When to use | Status |
|---|---|---|---|
| `lfcs-airtable-sync` | (programmatic) | Universal vault-MD → Airtable upsert adapter. **Every other lfcs-* skill calls this as final step.** Rocky should not invoke directly. Failure isolated — never blocks calling skill. | live |
| `chat-bootstrap` | Auto at start of mobile/Chat session, or explicit "bootstrap chat" | Generates current vault snapshot as a single copy-paste block for Rocky's phone Claude. Bridges stale-memory gap on mobile side. | live |

**Vault-side companion docs:** `11_OpsMan_LFCS/_Brain/CLAUDE.md`, `11_OpsMan_LFCS/Operations/skills/_README.md` (LFCS-scoped catalog with template + regression fixture rule), `11_OpsMan_LFCS/Operations/skills/_Log.md` (per-skill version history), `11_OpsMan_LFCS/Operations/skills/lfcs-standards/folder-spec.md` (canonical 11-section structure).

**Drift rule:** when adding/retiring an LFCS skill, update **both** this file's row AND the LFCS `_README.md` + `_Log.md`. This file is the cross-lane index; `_README.md` is the LFCS-detail catalog. They must agree on the skill's existence; details (Airtable behaviour, regression fixture, version history) live only in the LFCS catalog.

---

## Lane: 12_SiteOps (site-specific operations)

Vault-scoped skills under `HQ-Vault/12_SiteOps/Operations/skills/`. Lane scope: site-specific operational artifacts that aren't bid-side (LFCS) or growth-side (RateRight) — induction packs, ITPs, pre-pour checklists, toolbox-talk records, defect templates.

| Skill | Trigger | When to use | Status |
|---|---|---|---|
| `siteops-induction-pack` | "Build induction pack for [site] [crew]" / "Set up site induction" | Assemble standardised induction pack: white-card check, SWMS sign-on, site-specific brief, contact card, emergency procedure. Output to job's `05c - Inductions/`. | live |
| `siteops-itp-generate` | "Generate ITP for [pour-id/element]" / "ITP — [job] [zone]" | Generate Inspection & Test Plan markdown from TfNSW R-series spec hold-points + element taxonomy. Output to job's `07c - Inspection Records/`. | live |

**Vault-side companion docs:** `12_SiteOps/_Brain/CLAUDE.md`, `12_SiteOps/Operations/skills/_README.md` (catalog), `12_SiteOps/Operations/skills/_Log.md` (per-skill version history).

**When to add another:** if a site-ops sequence repeats 2+ times, codify it. Same drift rule as LFCS — update SKILLS.md row + lane `_README.md` + `_Log.md` together.

---

## Cross-lane / system skills

Plugin and harness-level skills available globally via the `Skill` tool. Listed for completeness — full inventory in the session's available-skills system reminder.

| Skill | Trigger | When to use |
|---|---|---|
| `caveman` | "caveman mode" / `/caveman lite\|full\|ultra` | Compressed output mode (~75% fewer tokens). Lite is HQ-validated for plan-mode design docs. |
| `caveman-commit` | "/commit" / "write a commit" | Conventional Commits commit message generator. |
| `caveman-review` | "/review" / "review this PR" | Compressed code-review comments (one line per finding). |
| `simplify` | (post-edit) | Review changed code for reuse, quality, efficiency; fix issues. |
| `loop` | "/loop [interval] [task]" | Run a prompt on a recurring interval, or self-paced. |
| `schedule` | "schedule a routine" | Create cron-driven remote agents. |
| `feature-dev` | "/feature-dev" | Guided feature dev with codebase understanding. |
| `frontend-design` | "/frontend-design" | Distinctive, polished frontend code. |
| `code-review:code-review` | "/review-pr" | Multi-agent PR review. |
| `pr-review-toolkit:review-pr` | "/review-pr" | Specialised PR review pipeline. |
| `claude-md-management:revise-claude-md` | "/revise-claude-md" | Update CLAUDE.md from session learnings. |
| `commit-commands:commit-push-pr` | "/commit-push-pr" | Commit + push + open PR in one go. |

Full set (incl. `update-config`, `keybindings-help`, `fewer-permission-prompts`, `app-audit`, `claude-api`, `agent-sdk-dev:new-sdk-app`, `claude-code-setup:claude-automation-recommender`, `hookify:*`, `ralph-loop:*`) appears in the available-skills list at session start.

---

## How to add a skill

1. Create `<lane-path>/skills/<skill-name>/SKILL.md` with frontmatter `name`, `description`, optional `user_invocable: true`.
2. Skill body must include sections: `## Purpose`, `## Trigger`, `## Behaviour`, `## Output`.
3. If the skill writes to vault and Airtable: call `lfcs-airtable-sync` as final step (LFCS lane) or document the equivalent sync for other lanes.
4. **Append a row to this file** in the right lane's table.
5. LFCS skills only: append version row to `11_OpsMan_LFCS/Operations/skills/_Log.md`. Cross-lane audit lives in `git log` — no separate audit file.

## How to retire a skill

1. Move `SKILL.md` to `HQ-Vault/_Archive/skills/<skill-name>/` with a top-line `**ARCHIVED YYYY-MM-DD — reason.**`.
2. Mark its row in this file with `~~strikethrough~~` + status `archived`. Don't delete the row.
3. LFCS skills only: append retirement row to `11_OpsMan_LFCS/Operations/skills/_Log.md`. Cross-lane audit lives in `git log`.
