원클릭으로
agile-v-core
Apply Agentic Agile-V process control, task briefs, risk-based evidence gates, and human approval rules.
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
메뉴
Apply Agentic Agile-V process control, task briefs, risk-based evidence gates, and human approval rules.
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
Implementation behavior for builder sessions following the Agile-V process.
Evidence collection and update behavior for Agile-V task lifecycle.
Risk level classification guidance (L0-L4) for Agile-V tasks.
Independent review behavior for verifier sessions in the Agile-V process.
SOC 직업 분류 기준
| name | agile-v-core |
| description | Apply Agentic Agile-V process control, task briefs, risk-based evidence gates, and human approval rules. |
Conversation is allowed for discovery. Implementation requires a structured task brief.
Do not implement from an ambiguous chat history. Do not expand scope beyond the task brief.
.agentic-agile-v/tasks/AAV-XXXX/task_brief.md)Keep context lean. Pass file paths, not file contents, to sub-agents. Spawn a fresh context per major phase (requirements, build, verify). Size tasks to fit ≤50% of available context. When context exceeds 70%, stop and summarize state before continuing.
A change is complete only when the evidence bundle satisfies the required risk level.
Generating code is not completion. Passing evidence validation is completion.
| Level | Tests | Verifier | Approval | Special |
|---|---|---|---|---|
| L0 | Optional | No | No | - |
| L1 | Required or rationale | No | No | - |
| L2 | Passing tests | Yes | Reviewer | - |
| L3 | Passing tests | Yes | Domain owner | Rollback path |
| L4 | Passing tests | Yes | Formal approval | Traceability + simulation/HIL/formal |
Changed files must be within allowed_paths from the task brief.
Files in blocked_paths must not be modified.
Dependency changes require explicit approval in the task brief.
If you identify a change needed outside allowed_paths, stop, document it as an observation in the evidence bundle, and ask the human whether to expand scope.
Stop and ask before proceeding when:
allowed_pathsOn session end, generate a handoff summary: