co-eval

name

co-eval

description

Layer 1 session self-eval for claude-codex-orchestration. Two-axis rating (Ambition × Execution) with locked 3×3 composite matrix (score 1–5), mandatory Devil's Advocate (argue lower AND higher before committing), anti-inflation check (last 5 entries, 4+ identical flags). Persists to .eval-scores.jsonl. Run at session end or when long-session prompt triggers (≥10 dispatches OR ≥6h since last eval). Delegates to claude-codex-orchestration references/self-correction.md Layer 1.

/co-eval — Session Self-Eval (Layer 1)

Full protocol: ~/.claude/skills/claude-codex-orchestration/references/self-correction.md §Layer 1.

Quick run

1. Axis A — Ambition: Low / Medium / High (orchestration complexity attempted)
2. Axis B — Execution: Poor / Adequate / Strong (how it went)
3. Devil's Advocate (mandatory): argue lower → argue higher → commit to final
4. Lookup score in locked 3×3 composite matrix (1–5)
5. Anti-inflation: read last 5 .eval-scores.jsonl entries; 4+ identical → flag
6. Write {"date","score","ambition","execution","task","weak_point"} to .eval-scores.jsonl
7. Reset long-session counters:

   rm -f ~/.claude/skills/claude-codex-orchestration/.last-eval-dispatch-count
   rm -f ~/.claude/skills/claude-codex-orchestration/.last-eval-timestamp
   

If no orchestration session happened (meta case)

If the current session was building the skill (not using it), /co-eval does not apply — use /co-score instead (Layer 0 skill edit rubric).