catalog

Catalog

Renders the in-repo claude-config inventory (rules, skills, agents, templates) from docs/catalog.md, grouped by purpose. Purely informational — no side effects.

The model does not auto-invoke this skill. It runs only when the user explicitly types /catalog (or one of its argument forms).

Arguments

• (no args) — Render all groups: rules, pipeline skills, leader toolkit, ops skills, agents, templates, ecosystem.
• skills — Skills only (pipeline + leader + ops).
• agents — Agents only.
• rules — Rules only.
• leader — Senior eng leader toolkit only (/onboard, /strategy-doc, /swot, /stakeholder-map, /1on1-prep, /present).
• templates — Templates only (drop-in scaffolds shipped by the repo).
• ecosystem — Third-party plugins this config plays well with.
• --mine — List skills present under ~/.claude/skills/ that are NOT documented in the canonical docs/catalog.md. Useful for finding plugin skills (caveman, superpowers, harness) and user-added skills outside the claude-config repo.

Arguments are mutually exclusive except --mine, which combines with any group filter.

Workflow

Step 1: Locate the catalog source

Resolve the canonical catalog path in this order:

1. <repo-root>/docs/catalog.md — if the current working directory is inside the claude-config repo.
2. ~/repos/claude-config/docs/catalog.md — the user's canonical clone.
3. If neither resolves, ask the user for the path.

The catalog file is the single source of truth. Do not re-derive tables from disk scans — render what's documented.

Step 2: Parse sections

docs/catalog.md uses H2/H3 headers and GitHub-flavored markdown tables. Map argument → section:

Argument	Section header(s) to render
(no args)	All H2/H3 sections in order
rules	## Rules (always active)
skills	### Pipeline + design-thinking, ### Senior eng leader toolkit, ### Other operational skills
agents	## Agents (specialized reviewers)
leader	### Senior eng leader toolkit
templates	## Templates
ecosystem	## Ecosystem (third-party plugins this config plays well with)

Render the matching tables verbatim. Do not paraphrase rows — the catalog file owns the descriptions.

Step 3: --mine flag handling

When --mine is present:

1. List directories under ~/.claude/skills/ via ls or Glob.
2. Extract documented skill names from docs/catalog.md (scan rows in the skills tables; strip leading / from slash-command names).
3. Report the set difference: ~/.claude/skills/* MINUS catalog-documented skills.
4. Group output by likely source: plugin name prefix (e.g., superpowers:*, caveman:*, harness:*) vs. unprefixed entries (likely user-added).

--mine is a diff, not a full inventory. Output explicitly labels what's NOT in the canonical catalog.

Step 4: Render

• Preserve the markdown table format from docs/catalog.md.
• Include the workflow mermaid diagram only when no args (rendering everything).
• For leader arg, prepend a one-line context: "Senior eng leader toolkit — 90-day ramp orchestration."
• For --mine with no extras found: "No skills found outside the canonical catalog."

Out of scope

• Interactive TUI / picker — output is markdown, not a menu.
• Real-time skill usage stats — no telemetry.
• Plugin discovery beyond ~/.claude/skills/ listing — does not query MCP servers or plugin marketplaces.
• Editing the catalog — read-only against docs/catalog.md.

Examples

/catalog

→ Full inventory (workflow diagram + rules + all skills + agents + templates + ecosystem).

/catalog leader

→ Senior eng leader toolkit table only.

/catalog skills --mine

→ Skills directories present under ~/.claude/skills/ but absent from docs/catalog.md, grouped by plugin prefix.

Related

• docs/catalog.md — canonical inventory this skill renders.
• rules/README.md — rule install contract; rules listed in the catalog are the HARD-GATE set.
• bin/link-config.fish — installs the symlinks that make catalog entries available at runtime.