메뉴
Use when inspecting or explaining Qingcheng AI chat complexity classification, prompt-template injection, SupervisorAgent routing, selected tool allowlists, or agent planning for a user message.
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
Use when inspecting AI assistant file-based long-term memory, session JSONL archives, soft consolidation, Dream updates, or assistant context isolation.
Use when running Agent golden-case evaluation, checking intent accuracy, tool-call accuracy, RAG grounding, job matching, interview follow-up policy, or resume rubric compliance.
Use when reading saved jobs, application status counts, manual follow-up notes, or the user's application tracking state.
Use when inspecting FastAPI routes, controllers, backend service boundaries, repository-backed endpoints, API surface, or route availability.
Use when changing or reviewing the Next.js frontend UI, Tailwind layout, chat/interview pages, resume/job/application screens, or frontend API wrappers.
Use when inspecting simulated interview Agent short-term session state, job-resume binding, question plan progress, answer signals, difficulty, follow-up strategy, or remaining focus.
SOC 직업 분류 기준
| name | chat-routing-tool |
| description | Use when inspecting or explaining Qingcheng AI chat complexity classification, prompt-template injection, SupervisorAgent routing, selected tool allowlists, or agent planning for a user message. |
Use this skill to inspect how the AI assistant would route a message. The runtime first classifies the turn as simple_answer or agentic_task:
simple_answer: direct model answer, no tool execution, no planning.agentic_task: SupervisorAgent planning, service-side tool allowlist, fact reading, generation, validation, and memory archival.scheduled_task: handled before normal model generation when the user asks to remind, repeat, list, pause, resume, or cancel scheduled tasks.帮我搜一下美团开发岗, 查一下腾讯 Java 岗, or 推荐北京后端实习 must route to agentic_task -> job_search.Prompt text is loaded through backend/app/prompts/registry.py from backend/app/prompts/templates/. Do not expose rendered prompts or hidden reasoning.
Run inside the API container:
docker compose exec api python /app/skills/chat-routing-tool/scripts/plan_turn.py --message "帮我找北京 Java 后端实习并结合简历分析"
Inputs: --message, optional --history-json, optional --tool-context-json.
The script emits compact JSON: ok, tool, input, result.complexity, result.intent, result.steps, result.tools, result.prompt_template_id, result.prompt_template_version, result.prompt_chars, and result.system_prompt_chars. Errors return ok=false with code and error.
Summarize whether the turn is simple or agentic, then summarize detected intent and selected tools. If knowledge_search appears, explain that it is used for Java/backend/八股 technical context. Never quote the rendered prompt.
python skills/chat-routing-tool/scripts/plan_turn.py --self-test
python skills/chat-routing-tool/scripts/plan_turn.py --help
docker compose exec api python /app/skills/chat-routing-tool/scripts/plan_turn.py --message "你好"
docker compose exec api python /app/skills/chat-routing-tool/scripts/plan_turn.py --message "帮我找北京 Java 后端实习并结合简历分析"
docker compose exec api python /app/skills/chat-routing-tool/scripts/plan_turn.py --message "帮我搜一下美团开发岗"
docker compose exec api pytest tests/scheduled_tasks -q