ワンクリックで
self-improvement-governance
Governance rules for bounded self-improvement candidates, promotion gates, and rollback policy.
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
メニュー
Governance rules for bounded self-improvement candidates, promotion gates, and rollback policy.
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
SOC 職業分類に基づく
Git Worktree: parallel working trees for isolated branch-level execution, debugging, and safe experimentation.
Durable project-memory rules for `.agent-memory/` plus session-memory boundaries.
Planning-track selection, epic/feature decomposition, readiness gates, and plan-delta rules.
Shared contract for independent review agents and durable review findings.
Review routing, independent post-implementation review gates, multi-model escalation, and targeted optimization follow-up.
Independent evaluation protocol for bounded self-improvement candidate proposals.
| name | self-improvement-governance |
| description | Governance rules for bounded self-improvement candidates, promotion gates, and rollback policy. |
| license | See repository LICENSE |
| user-invocable | false |
Use this skill for any workflow that proposes, evaluates, or promotes bounded self-improvement candidates.
Only the dedicated low-risk overlay namespace is auto-mutable:
.agents/skills/self-improvement-overlays/**All other paths are protected by default.
Do not auto-promote changes that:
opencode.json.opencode/agents/*.md except through explicit human approvalAuto-promotion requires:
Otherwise, the candidate must remain NEEDS_HUMAN or REJECT.
Rollback to the last accepted baseline when:
The rollback must restore the prior baseline and quarantine the rejected candidate.