From odin-loop
Odin-Loop's memory raven. Analyzes your Odin-Loop run history and raw Claude Code session transcripts to find where you skip stages, re-work, or loop back, then proposes concrete edits to a loop's YAML. Use for `/odin refine`, or when the user asks to improve/tune/refine a loop based on past sessions or history.
How this skill is triggered — by the user, by Claude, or both
Slash command
/odin-loop:muninnThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
Muninn ("memory") is the raven that flies over your past work and reports back.
Muninn ("memory") is the raven that flies over your past work and reports back. This skill closes Odin-Loop's outer loop: observe how you actually work → propose edits to your loop definition → you approve → the loop improves.
Triggered by /odin refine [loop-name]. If no loop name is given, default to the
active run's loop, else spec-harness-tdd.
Gather signals (cheap, deterministic). Run the bundled analyzer — do NOT read raw transcripts into context (they can be megabytes):
python3 "${CLAUDE_PLUGIN_ROOT}/scripts/analyze_sessions.py" --cwd "$(pwd)"
It emits compact JSON: Odin-Loop run history (loopbacks, gate failures,
abandoned runs) plus raw-session heuristics (tool usage, churn_ratio,
avg_turns_before_first_edit, test_runs, heavily-reworked files). It reports
aggregates only — never message content.
Read the target loop YAML (.odin-loop/loops/<name>.yaml or the built-in
loops/<name>.yaml) so proposals reference real stage ids and gates.
Interpret signals → proposals. Map signals to concrete, minimal edits. Prefer tightening an earlier gate over adding stages. Honest mapping:
| Signal | Likely cause | Proposed edit |
|---|---|---|
most_looped_stage = implement (high loopbacks) | spec/harness too weak; rework downstream | strengthen the interview or harness-design gate check; don't just raise max_iterations |
high churn_ratio (>2.5) + test_runs ≈ 0 | building without a verifiable target | recommend adopting/strengthening the harness stages; flag missing test discipline |
high avg_turns_before_first_edit | lots of deliberation before code | the interview stage already captures this — confirm it's used; or add an explicit plan stage |
runs abandoned at interview | interview too long / fatiguing | add a question cap hint to the interview prompt; split into must-have vs nice-to-have |
a gate always passes first try (no loopbacks, all ai) | gate may be rubber-stamping | suggest a stricter check, or flip to ai+human for that stage |
gate_failures_by_stage concentrated on one stage | that stage's check is hard to satisfy as written | clarify the check or split the stage |
If found_run_history is false (no Odin-Loop runs yet), rely on raw-session
heuristics only and SAY SO — they are softer signals.
Write a proposal, do not auto-apply. Save a refinement report to
.odin-loop/refinements/<timestamp>.md (use date for the stamp) containing:
🦅 무닌 제안 N건. 적용하려면
/odin refine apply, 개별 선택/수정은 말씀해 주세요.
Apply only on approval. When approved, edit the loop YAML in place
(bump its version), and note in the report which edits were applied.
npx claudepluginhub choo121600/odin-loop --plugin odin-loopGuides creation, editing, and verification of skills for AI coding agents using test-driven development with subagent scenarios. Use when authoring or debugging skills.