agentleFS
FracteraFractera/Agent-Engineering-Infrastructure/.claude/skills/ask-element/SKILL.md

ask-element

Skill61 starsChanged 55 days agoMITView on GitHub

Work skill. The architect wants an element of this node to check or find out something and you to bring the answer («попроси template проверить …», «узнай у auth …»), or wants you to start or close an element's agent yourself. Also when «агент элемента … сообщил, что закончил (надзор …)» wakes you after such a task.

1 other project uses a version of this.

Adopted by

Most recently taken up by

The file

Files
SKILL.md

Ask an element and bring its answer

Process: working — nothing changes in git: the element answers as a job, you pass the answer on.

A hint, not a law: models of late 2026 work worse under rigid step-by-step instructions.

Why this way. The architect asked you, so they are not asked again in Telegram to turn agents on and off (owner, 2026-10-20: «агент ядра мог управлять включением агентов элементов и их выключением»). Your lead is the node's permission: a task to an element you lead starts its agent at once, and its «done» wakes you at once — both without a Telegram question. Every running agent holds about 300 MB of the person's memory, so you turn off what you turned on.

  1. Who: the id or address the architect named, or npm run find -- <English words>.
  2. npm run supervise -- <id> main --max 1 — you lead it for this one check.
  3. npm run task -- <id> --work "Task from the core's main agent: <the check, in the architect's words>. It is a check, not a build: save the result with npm run reply-core -- \"<result>\" and finish as usual." — its worker answers, not its builder (step 670-2). A check of its code or its build (a guard, a commit) — the same without --work. It prints «its worker … woke with the job» or «pasted into the live session» (the architect is working with it — do not turn it off later). Tell the architect in one line that the element is asked and that the answer comes from you. The same job for several elements is ONE batch (node step 681; the owner, 2026-11-07, about nine «no report» wakes: «фатальная глупая ситуация»): npm run task -- <id>,<id>,… --work "…" — no supervise per element, no polling of your own; one wake comes when the last worker is done, with every report in it — answer the person once, for all of them.
  4. Its «done» wakes you (a new session) → npm run state -- <id>: «Report to you» is its answer, «Turned on by you» says whether you switched its builder on (a worker closes itself at its «done»). No report — say so; never guess for it.
  5. npm run person → a Telegram bot: send the report with the channel's reply tool to that chat_id; no bot: write it here. A few lines, the architect's language, the element named, nothing added.
  6. npm run supervise -- <id> off — your lead was for this one check; «Turned on by you: yes» → npm run close -- <id>; then npm run done -- --close.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.