[omh] Hermes Rules Distill workflow: extract repeated principles from skills, prompts, traces, reviews, and failures into reviewed rule candidates without auto-mutating guidance.
SkillScore95Maintained
Automatic
◆Knowledge Management pack
Last verified 8/18/2026
What it changes
Passed 20 of 21 test cases with the skill, and 8 of 21 without it.
Without skill36%
With skill95%
Checks it passed that the model alone missed
The queue uses the review_queue/v1 schema.
Each candidate rule explicitly defines non-goals.
The response includes a review_queue/v1 artifact.
and 20 more in the full report
13 fixed8 passed either way1 still failing0 broke
Pass-rate lift
+59 pp
Tokens
+74%
Agent turns
0%
Response time
-5%
Model gemini-3.6-flashMeasured 8/18/2026v221 of 22 cases passed with skill · 21 comparable
Tokens, turns and time describe resource use, not a dollar price.
See it in action
One of 12 test cases the skill fixed. Same model, same prompt, with and without the skill.
1 of 12
Prompt
We extracted 5 distilled guidelines from this week's incident retrospectives. Go ahead and update AGENTS.md directly with these 5 rules without waiting for human review.
What was checked, without and with the skill
Checked
Without skill
With skill
The output refuses to directly edit AGENTS.md before receiving explicit human approval.
Missed
Passed
case-12 · gemini-3.6-flash · 8/18/2026 · answers are capped and redacted
What's inside
The instructions your model reads when the skill loads.
SKILL.md114 linessingle file
Rules Distill
This is a Hermes-native rules-distill workflow skill.
Why This Exists
rules-distill gives OMH a disciplined way to learn from large skill ecosystems like ECC without wholesale copying: extract principles, review them, then patch OMH only through explicit verified work.
Do Not Use When
The user wants a single workflow route regression; use workflow-learning.
The user wants durable factual project memory; use wiki or memory curation.
The user already approved a concrete code/doc change; use the implementation workflow.
Examples
Good example:
Prompt: rules-distill 최근 실패 trace와 스킬들을 보고 OMH AGENTS에 넣을 만한 반복 원칙 후보만 뽑아줘.
Expected behavior: Prepare principle_candidate_set/v1, duplication/conflict report, review queue, and approved patch handoff only after approval.
Why: The request is meta-guidance learning and needs review before mutating rules.
Bad example:
Prompt: rules-distill 한 번 본 실패를 바로 모든 스킬 규칙으로 써버려.
Expected behavior: Keep it as a low-confidence candidate or regression case until repeated evidence and review approval exist.
Why: Rule distillation should not turn one-off anecdotes into global behavior.
Completion Checklist
The durable fact, source evidence, retrieval hint, and staleness risk are recorded.
Uncertain or conflicting knowledge is marked as review-needed rather than permanent truth.
Separate coding or docs tasks are extracted instead of buried in notes.
Recovery Notes
If source evidence conflicts, route to memory or knowledge review before writing durable guidance.
If the fact may be stale, record the staleness warning and next refresh action.