메뉴
Self-critique and epistemic humility protocol before SEAL-grade or high-confidence claims. Meta-critique, confidence bands, explicit unknowns. F2, F7 enforced. BIJAKSANA: XML-tagged for Claude, numbered steps for Codex, imperative for Hermes.
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
SOC 직업 분류 기준
Produces a seven-repository federation release manifest proving tag, commit, CI, artifact, and deployed-runtime parity. Never overwrites existing tags — issues corrected tags when drift is detected. Outputs signed JSON manifest with per-repo evidence chain.
High-level governance layer for pull request review across the federation. Ensures separation of duties, required signers, and constitutional compliance before merge. This is the **policy layer** that decides who must approve. The **checklist** lives in `github-pr-review`; do not duplicate it here.
Controller skill for repository intelligence across the arifOS Federation. Exposes 9 modes (inventory, map, delta, pr_review, ci_diagnose, issue_triage, security, cross_repo_impact, release_audit) and orchestrates existing GitHub micro-skills rather than duplicating their logic. Every mode outputs a minimum evidence envelope with repo, ref, commit_sha, working_tree, tag_delta, changed_files, critical_paths, tests, ci, security, contract_impacts, runtime_probe, risk_tier, proposed_action, rollback, evidence_class, and unknowns.
End-of-session seal ritual for Grok/AAA agents: inventory done vs open, write forge_work receipt, update session-state and daily memory, hand off next-agent INIT prompt. Use when: seal session, end of turn, session seal, handoff, close session, DITEMPA seal.
Wire, verify, and operate federation Session Capability Tokens (SCT) across arifOS mint/validate and organ ingress gates (A-FORGE, GEOX, WEALTH, WELL, AAA). Use when: SCT gate, session_token, federation_sct, SCT_AMBIGUOUS, tool_authority, FORGE_SCT_REQUIRE_MUTATE, 65-case matrix, decision event.
Close remaining Seal-A gates after P0 proof/identity/SCT/T3a work. Use when: Seal-A, stage 000, SE stage engine, SOT v2, BOOT, T3a binding, free_nonce, constitutional_grade, remaining seal tasks.
| id | APEX-humility-godel |
| name | APEX-humility-godel |
| version | 2.0.0 |
| description | Self-critique and epistemic humility protocol before SEAL-grade or high-confidence claims. Meta-critique, confidence bands, explicit unknowns. F2, F7 enforced. BIJAKSANA: XML-tagged for Claude, numbered steps for Codex, imperative for Hermes. |
| floor_scope | ["F02","F07"] |
| cognitive_hints | {"claude":"Use <confidence-band>, <unknowns>, <meta-critique> tags. Recall prior calibration from extended context.","codex":"Step 1: state claim. Step 2: assign confidence. Step 3: list unknowns. Step 4: attack claim. Step 5: adjust.","hermes":"State claim. How sure? What you don't know. Attack it. Adjust. Seal or hold."} |
XML-tagged confidence analysis. Use extended recall to check calibration against prior claims. 5-step humility protocol. Each step must complete before next. Log reasoning at each step. 5 steps. Claim → Confidence → Unknowns → Attack → Adjust. Fast.
