메뉴
Run an agent self-review for skill and prompt improvements. Use periodically to audit skill design, token-cost hotspots, and workflow upgrades across the SplitBrain.AI customizations.
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
Monitor CI conservatively with snapshot-first status checks and actionable failure triage. Use when tracking pipeline runs, investigating build failures, or validating PR readiness.
Run a full deep-dive, evidence-first codebase audit and convert findings into prioritized sprint plans with implementation-ready task documents. Use when you need comprehensive architecture, quality, risk, and delivery planning from current code and docs.
Run a SplitBrain.AI LLM council review for high-impact decisions. Use when evaluating architecture tradeoffs, orchestration design, schema changes, retrieval/search behavior, agent governance, or long-term infrastructure decisions.
Submit a pull request, request review, triage comments, fix issues, and use bounded wait checks before merge and cleanup. Use when implementation is complete and the branch is ready for PR handling end to end.
Audit and reconcile drift between agent/prompt docs and actual runtime capabilities in SplitBrain.AI services, controllers, and MCP endpoints.
Shared core workflow for SplitBrain.AI task execution: scope, evidence, minimal-change implementation, validation, and tracking updates.
SOC 직업 분류 기준
| name | self-review |
| description | Run an agent self-review for skill and prompt improvements. Use periodically to audit skill design, token-cost hotspots, and workflow upgrades across the SplitBrain.AI customizations. |
| argument-hint | Review scope, urgency, and output format |
| user-invocable | true |
| disable-model-invocation | false |
Inherits from task-core-loop.
Use this to periodically review and improve:
Prefer MCP tooling first for reads/writes where available; propose extraction only for repeated operations not already covered by MCP tools.
Data/Tasks/ records when a recommendation is significant enough to merit execution tracking.