القائمة
Use when creating, listing, testing, or debugging Qingcheng AI scheduled tasks, task inbox results, worker execution, natural-language reminder parsing, or Telegram task commands.
التثبيت باستخدام Codex أو Claude انسخ هذا Prompt والصقه في Codex أو Claude أو مساعد آخر ليراجع صفحة Skill ويثبّتها لك.
استنادا إلى تصنيف SOC المهني
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 inspecting or explaining Qingcheng AI chat complexity classification, prompt-template injection, SupervisorAgent routing, selected tool allowlists, or agent planning for a user message.
Use when changing or reviewing the Next.js frontend UI, Tailwind layout, chat/interview pages, resume/job/application screens, or frontend API wrappers.
| name | scheduled-task-tool |
| description | Use when creating, listing, testing, or debugging Qingcheng AI scheduled tasks, task inbox results, worker execution, natural-language reminder parsing, or Telegram task commands. |
AI assistant scheduled tasks are application-level jobs stored in PostgreSQL, not OS cron files. The scheduler is run by notification-worker through app.tasks.telegram_tasks.run_scheduled_tasks_once().
Supported schedule types:
once: one-time reminders such as 明天上午 9 点提醒我检查投递状态.interval: repeated intervals such as 每 30 分钟检查一次岗位.cron: five-field cron expressions and common workday phrases such as 工作日早上 9 点提醒我看投递.一分钟后提醒我看岗位, 半小时后提醒我检查简历, and 每周一早上 9 点提醒我检查投递.Safety boundary:
chat_answer, resume_profile, job_search, application_list, knowledge_search, telegram_send.assistant_task_inbox; Telegram-created tasks also send results back to Telegram.source_channel=telegram, delivery_channel=telegram, and telegram_chat_id; execution results are sent to the same Telegram chat and also copied to assistant_task_inbox.This skill has no application-data Python script. Use backend tests, API checks, and worker functions for validation.
Run unit tests from the host:
python -m pytest backend/tests/scheduled_tasks -q
Run inside Docker after migrations:
docker compose exec api pytest tests/scheduled_tasks -q
docker compose exec api python -c "from app.tasks.telegram_tasks import run_scheduled_tasks_once; print(run_scheduled_tasks_once())"
API checks:
curl http://localhost/api/v1/scheduled-tasks -H "Authorization: Bearer <access_token>"
curl http://localhost/api/v1/task-inbox -H "Authorization: Bearer <access_token>"
Telegram commands:
/tasks
/cancel_task <id-prefix>
/pause_task <id-prefix>
/resume_task <id-prefix>
Task API items include task_id, title, instruction, status, schedule_type, schedule_label, next_run_at, next_run_at_local, delivery_channel, and last_error.
Inbox API items include inbox_id, task_id, task_run_id, title, content, status, created_at, and safe metadata summaries.
Chat SSE end.metadata includes scheduled_task_action, scheduled_task_id, schedule_summary, next_run_at, task_inbox_id, and tool_calls_summary.
When explaining scheduled tasks, say clearly that tasks are persistent, worker-driven, auditable, and scoped to the AI assistant. Do not describe them as operating-system cron or as arbitrary automation with full machine access.
python -m compileall -q backend/app
python -m pytest backend/tests/scheduled_tasks -q
cd frontend
npm test -- scheduled-task-panel
npm run typecheck