Command Claude Code
Author a launch guide into guides/ — grounded in a methodology corpus you ingest (optional) + the existing guides, in the right shape (walkthrough/checklist/notice), carrying the guide-anchor contract so /guides:integrate can place it in the bootstrap pipeline.
Command Claude Code
Design and create a new hook from a description.
Command Claude Code
Temporarily disable a hook by moving it from settings.json into a disabledhooks section, with a one-step path to re-enable later.
Command Claude Code
Analyze friction points — find patterns that suggest missing hooks.
Command Claude Code
Test all hooks with synthetic payloads and measure execution time.
Command Claude Code
Mark a recurring system issue resolved with a permanent fix summary.
Command Claude Code
Karpathy autoresearch loop — plan a closed-loop experiment, review, then run autonomously in an isolated worktree. Optimize any editable artifact (agent spec, skill, hook policy) against a scalar metric.
Command Claude Code
Ingest external knowledge from files, links, or YouTube videos and apply learnings to the system.
Command Claude Code
Migrate the manifest to a target MC version. Dry-run by default; --apply to write.
Command Claude Code
Print .claude/manifest.json (pretty by default, --json for compact).
Command Claude Code
Validate the current .claude/manifest.json against the v1 manifest schema and report any drift, missing fields, or schema violations.
Command Claude Code
Registry of all maps — shows every map, its source, last updated, and staleness.
Command Claude Code
App structure — routes, components, libs, how they connect.
Command Claude Code
Enforcement coverage — hooks, gates, gap analysis, open/closed gaps.
Command Claude Code
Hook wiring diagram — events, matchers, scripts, execution order.
Command Claude Code
Memory store relationships — who reads/writes each store, entry counts.
Command Claude Code
Require/touchpoint graph — static require edges over scripts/, plus paths-consumer, hook-wiring and skill-call edges, with dynamic require sites flagged unresolved. Feeds the lean gauntlet's touchpoint selector.
Command Claude Code
Skill dependency graph — namespaces, cross-references, data flow.
Command Claude Code
Regenerate step tables in canonical docs from requirements/00-canonical/STEPS.json — one region per phase the product declares.
Command Claude Code
Tool registry — skills, hooks, external CLIs, API services, npm scripts, platform tools.
Command Claude Code
Compare your MC installation against the latest version — find stale, new, and missing items.
Command Claude Code
Create a guarded MC deprecation proposal for an agent, skill, hook, path, requirement, pattern, or generated file.
Command Claude Code
Diff canonical MC against an installed product — version/staleness, framework-file drift (stale vs locally-modified), coverage gaps, and skills/agents/hooks delta. Read-only; reports on the product, never edits it.
Command Claude Code
Unified MC diagnostic — runs every health check in one place. Like /mc:health but full-coverage.
At most 3 mods per repository are shown here, and a mod shipped inside a plugin is left to that plugin's page — the rest are on their repository pages: