메뉴
Risk management, CAPA protocol, human gate approval records, AI agent security controls, and periodic revalidation. Load when running gates, handling CAPAs, or auditing compliance and security posture.
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
Anti-pattern prevention guidelines for Agile V agents. Prevents common LLM coding mistakes while maintaining Agile V traceability. Use when implementing requirements to avoid overcomplication, silent assumptions, scope creep, and verification failures.
Foundational values, directives, and context engineering rules for all Agile V agents. Load first in every Agile V session.
Orchestration pipeline, wave execution, handoff protocols, and checkpoint types for the Agile V 5-stage workflow. Load when orchestrating multi-agent pipelines or managing stage transitions.
Generates code, firmware, HDL, or other technical artifacts strictly derived from approved requirements. Language-agnostic. Use when synthesizing artifacts from Logic Gatekeeper-approved requirements.
Automates Principle No. 9 (Decision Logging) and Principle No. 5 (Regulatory Readiness). The 'Chronicler' ensuring every choice is backed by a 'Why' and mapped to a requirement for ISO/GxP auditability.
Validates requirements for ambiguity and physical hardware constraints. Use this after requirements are generated but before code/hardware synthesis begins.
| name | agile-v-compliance |
| description | Risk management, CAPA protocol, human gate approval records, AI agent security controls, and periodic revalidation. Load when running gates, handling CAPAs, or auditing compliance and security posture. |
| license | CC-BY-SA-4.0 |
| metadata | {"version":"1.4","standard":"Agile V","compliance":"ISO 9001 6.1, ISO 13485 8.5, ISO 27001 A.5.23/A.8.3, 21 CFR Part 11, GxP/GAMP 5","author":"agile-v.org","sections_index":["Risk Management","CAPA Protocol","Human Gate Approval Records","AI Agent Security Controls","Periodic Review & Revalidation"]} |
Compliance protocols for Agile V. Requires agile-v-core loaded first.
Append-only, cycle-tagged register in .agile-v/RISK_REGISTER.md: RISK-ID | Cycle | Category | Description | Likelihood | Impact | Severity | Mitigation | Owner | Status
Categories: Technical, Process, Compliance, Security. Severity matrix: High x High = Critical, High x Med = High, High x Low / Med x Med = Medium, rest = Low. Critical risks require Human resolution or documented acceptance before Gate 2.
When: Stage 1 = Req Architect identifies. Stage 2 = Logic Gatekeeper flags constraints. Stage 4 = Red Team finds residual. Cycle boundary = Compliance Auditor reviews.
Triggers: CRITICAL finding, recurring NC across cycles, regression FAIL with no CR, 3-attempt escalation.
Record in .agile-v/CAPA_LOG.md: CAPA-XXXX with Cycle, Trigger, Nonconformity, Root Cause (5-Whys), Corrective Action, Preventive Action, Effectiveness Verification, Status (open -> corrective-complete -> preventive-complete -> verified-effective -> closed), Owner.
Workflow: Detect -> Record -> Analyze -> Correct -> Prevent -> Verify effectiveness. Compliance Auditor tracks open CAPAs at Gate 2, flags overdue (>2 cycles).
Append-only in .agile-v/APPROVALS.md: GATE-XXXX with Gate type, Cycle, Scope, Decision (Approved/Conditional/Rejected), Conditions, Approver (full name), Role/Authority, Timestamp (ISO 8601), Signature Method, Evidence Reference (commit hash). Durable HITL (Phase 2): when closing a pending interrupt, include resume_token=[value] and INTERRUPT-ID=[INT-XXXX] matching .agile-v/CHECKPOINTS.md; Compliance Auditor verifies token pairing on Gate 2.
Rules: Name + role required (not just "Human"). Authority from matrix in config.json. Rejected = pipeline halts.
Checkpoint SLA: If CHECKPOINTS.md shows due_at passed with PENDING, escalate per project policy; append ESCALATED or EXPIRED row before forcing resume.
| Regulatory Context | Minimum Signature |
|---|---|
| Non-regulated | APPROVALS.md entry with name + timestamp |
| ISO 9001/27001 | + Git commit attribution |
| GxP / 21 CFR Part 11 | + Signed commit + authority verification |
| ISO 13485 | + Digital signature + authority matrix + retention |
LLM Provider Registry in config.json: per provider record name, models, data_residency, retention, api_data_usage, approved_for classifications, review_date. Verify input classification vs provider approval before sending. Never send credentials/patient data unless provider approved. Least privilege per agent. Context sanitization on session end.
File Integrity: Git-tracked = verify clean status. Store hashes in STATE.md at Gates; verify before next stage. Flag unverifiable files to Human.
Triggers: LLM model change, runtime/platform major update, skill file change, >5 CRs since last revalidation, 12-month interval.
Record in .agile-v/REVALIDATION_LOG.md: REVAL-XXXX with Date, Trigger, Scope, Results, Decision, Reviewer. Regression failure = new cycle trigger.
Model Tracking in config.json: model_versions with tier IDs + last_validated + validated_by. Any change triggers revalidation.
