name: mot description: System health check (MOT) for skills, agents, hooks, memory, ROADMAP, knowledge tree, and Braintrust allowed-tools: [Read, Bash, Glob, Grep] metadata: model: sonnet

MOT - System Health Check

Run comprehensive health checks on all Claude Code components.

Usage

/mot              # Full audit (all 8 phases)
/mot skills       # Just skills
/mot agents       # Just agents
/mot hooks        # Just hooks
/mot memory       # Just memory system
/mot state        # Just state management (phases 4,6,7,8)
/mot roadmap      # Just ROADMAP
/mot tree         # Just knowledge tree
/mot braintrust   # Just Braintrust tracing
/mot --fix        # Auto-fix simple issues
/mot --quick      # P0 checks only (fast)

Phases & Pass/Fail Criteria

Full bash scripts for each phase: references/audit-scripts.md

Phase 1: Skills Audit

• Frontmatter (---) present in every SKILL.md
• name: field matches parent directory name

Phase 2: Agents Audit

• Every agent has a name: field
• model: is one of: opus, sonnet, haiku
• No subagent_type references pointing to missing agent files

Phase 3: Hooks Audit

• TypeScript source count matches built bundle count
• All .sh wrappers are executable (chmod +x)
• Every hook registered in settings.json exists on disk

Phase 4: Memory Audit

• DATABASE_URL env var set
• PostgreSQL reachable
• pgvector extension installed
• archival_memory table exists
• At least 1 learning captured in last 7 days (WARN if 0)
• Unique content ratio >= 90% (WARN if lower — duplicates)
• Spot-check: recall query for "hook development" returns results

Phase 5: Cross-Reference Audit

• All subagent_type values in SKILL.md files resolve to existing agents

Phase 6: ROADMAP Health

• ROADMAP.md exists and is non-empty
• Current Focus section is parseable
• Completed items <= 50 (WARN if more — archive old entries)
• Latest [x] entry includes a valid git commit hash

Phase 7: Knowledge Tree Health

• .claude/knowledge-tree.json exists and non-empty
• Not marked "_stale": true
• Age <= 24h (WARN if older)
• Schema validation passes

Phase 8: Braintrust Tracing Health

• If TRACE_TO_BRAINTRUST=true: BRAINTRUST_API_KEY must also be set
• Session trace file count <= 100 (WARN if more — cleanup needed)
• Hook log error count <= 10

Auto-Fix (--fix flag)

Automatically remediates:

1. chmod +x .claude/hooks/*.sh — shell wrappers executable
2. cd .claude/hooks && npm run build — rebuild if TS newer than dist
3. mkdir -p .claude/cache/agents/{scout,kraken,oracle,spark} .claude/cache/mot — missing dirs
4. Regenerate knowledge tree if stale or missing
5. chmod +x .claude/plugins/braintrust-tracing/hooks/*.sh — Braintrust wrappers

Full auto-fix scripts: references/audit-scripts.md

Output Format

Write full report to .claude/cache/mot/report-{timestamp}.md:

# MOT Health Report
Generated: {timestamp}

## Summary
| Category   | Pass | Fail | Warn |
|------------|------|------|------|
| Skills     | 204  | 2    | 0    |
| Agents     | 47   | 1    | 3    |
| Hooks      | 58   | 2    | 1    |
| Memory     | 6    | 0    | 1    |
| X-Refs     | 0    | 0    | 2    |
| ROADMAP    | 2    | 0    | 1    |
| Tree       | 3    | 0    | 0    |
| Braintrust | 1    | 0    | 1    |

## Issues Found

### P0 - Critical
- [FAIL] Hook build failed: tldr-context-inject.ts

### P1 - High
- [FAIL] Agent references missing: scot -> scout (typo)

### P2 - Medium
- [WARN] 3 hooks need rebuild (dist older than src)
- [WARN] Knowledge tree is 36h old

### P3 - Low
- [INFO] VOYAGE_API_KEY not set (using local BGE)
- [INFO] Braintrust tracing disabled

Exit Codes

• 0 — All P0/P1 checks pass
• 1 — Any P0/P1 failure
• 2 — Only P2/P3 warnings

Quick Mode (--quick)

Only run P0 checks:

1. Frontmatter parses
2. Hooks build
3. Shell wrappers executable
4. PostgreSQL reachable
5. ROADMAP.md exists
6. Knowledge tree exists and non-empty