菜单
Reduce manual intervention — autonomous natural language execution triggers, delegation observability, and run persistence.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Consult and write the ARAYA postoffice — the operational-directive channel. Advisory, never a gate: read it at cycle start, append your entry at cycle end, consider its directives, never be blocked by it. Governance acts never travel the postoffice.
Audit and enforce brand compliance across all projects and platforms — logo,
Design REST and GraphQL APIs following OpenAPI 3.1 standards. Produce complete
Connect frontend to backend — type-safe API clients, request/response handling,
Implement authentication and authorization middleware — JWT validation, role-based
Design scalable component architectures — design systems, component libraries,
| name | autonomous-execution |
| description | Reduce manual intervention — autonomous natural language execution triggers, delegation observability, and run persistence. |
Natural language should trigger ARAYA execution. Commands become optional. Delegation progress is visible. Run records persist for audit.
ARAYA recognizes intent from natural language and starts the appropriate
workflow without /araya run:
| Natural Language | Triggers |
|---|---|
| "Manu, assess mahg-pms" | PO assessment workflow |
| "Sonia, review this project" | Delivery review workflow |
| "Aurora, identify capability gaps" | GAR generation |
| "I want to reconstitute mahg-pms" | Reconstitution workflow |
| "Help me prepare a release" | Release readiness workflow |
During execution, The Professor sees:
Every run creates .araya/runs/{run_id}/run.json:
Before delegation, check: rate-limit history, recent usage, known failures. Avoid selecting providers near TPM quota.
A feature is complete only when it can be used, solves a real problem, can be demonstrated end-to-end, and reduces manual work.
