agentleFS
Sign inStart free
ilya-kozyrevilya-kozyrev/delamain/skills/status/SKILL.md

status

Skill5 starsChanged 37 days agoMITView on GitHub

Answer questions about Delamain agents, stages and hubs, their tasks, owner questions, locks and idle stages. Use for “what is running”, “which agents are alive”, “task status”, “who needs an answer”, “что сейчас работает”, “какие агенты живы”, “что с задачами”, “кто ждёт ответа”, “какие блокировки” and “какие хабы можно закрыть”. Not for version-control working-tree state, build/CI state, system processes/ports, or tickets/issue trackers.

177 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use off, list and 10 more.

Other files people use alongside this one.

Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
  • StakeholderSenseiSkill

    off

    Disable polygraph for this repo (user-only escape hatch; all hooks no-op)

    Also in 4 projects that use this

  • SuShuhengSkill

    list

    Use when checking obshare-cli upload history, finding document tokens, or reviewing previous upload URLs and permission snapshots.

    Also in 8 projects that use this

  • kennethleungtySkill

    pause

    Pause — stop the music (alias for /stop)

    Also in 6 projects that use this

  • codebycoreySkill

    new

    Start a new change with a proposal. Use when starting a feature, creating a proposal, or beginning planned work.

    Also in 7 projects that use this

  • michaelenglandSkill

    import

    This skill should be used when the user wants to import an existing .pptx file into deckwright, bring in a PowerPoint that was created outside this tool, reverse-engineer a presentation for editing, or invoke /deckwright:import. Converts a .pptx into a PptxGenJS generation script so it can be edited and regenerated through the deckwright pipeline.

    Also in 5 projects that use this

  • genomewalkerSkill

    resume

    Resume a thread by loading its ~800-token context capsule

    Also in 15 projects that use this

The file

We split this skill into the rules it gives an agent

6 rules. 0 are common, 2 are distinctive, and 4 are very specific to this project.

  • 0

    Common

    Most of the 145 edited versions we read keep these.

  • 2

    Distinctive

    This author's own choices. Worth a look.

  • 4

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Use --stage <stage> for a named or clearly implied stage (repeatable), otherwise cover every stage.

    line 11
  2. Very specific:

    Use --all only when the user asks about finished or old work.

    line 12
  3. Distinctive:

    After a timeout, rerun with a longer one and wait for a fresh snapshot.

    line 15
  4. Distinctive:

    Read the fresh snapshot before answering.

    line 17
  5. Very specific:

    Use action.elapsed_s for action duration and started_at for run duration.

    line 19
  6. Very specific:

    Keep raw JSON and command names out of the answer unless requested.

    line 32

“Common” is measured: kept by at least 6 in 10 of the 145 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Running work

  1. Resolve the plugin root from PLUGIN_ROOT (Codex), CLAUDE_PLUGIN_ROOT (Claude), or this skill's canonical location (resolve symlinks first, then go two directories above skills/status). Run that root's bin/agent-top --json through the host's shell tool; an unrelated command may shadow it on PATH. Use --stage <stage> for a named or clearly implied stage (repeatable), otherwise cover every stage. Use --all only when the user asks about finished or old work; --json --agent <role> --feed 10 gives recent activity. On the owner's data, snapshots take 20–45 s, --all up to ~160 s. In Codex, wait through execution continuation tools. In Claude Code, set the shell tool timeout to at least 5 min, or run in background and wait for completion. After a timeout, rerun with a longer one and wait for a fresh snapshot; never use stale data or report unavailable merely because collection timed out.
  2. Read the fresh snapshot before answering. Group by stage: live agents with role, task/current action and duration, relevant finished/failed work since the previous snapshot in this conversation, owner questions and held locks. Use action.elapsed_s for action duration and started_at for run duration; age_s is time since the last log event, not runtime. Describe native sessions' activity-based liveness as approximate. An unread inbox is a message awaiting the agent, not an owner question. questions_ok: false means the question register is unavailable. Report missing data or non-timeout command failure as unavailable; an empty successful snapshot means no observed work. stage_info.<S>.idle marks an open stage with no live agent and no activity (own lines, logs, hub session; not [pace] or broadcasts) for idle_after_s: name such stages once as candidates to close (hub close, skill hub; the owner decides). closed_hidden lists closed stages left out of the snapshot: they run nothing; mention them only when asked about old or closed work. desktop_idle: idle Desktop sessions holding memory (archiving: skill hub). A hub lives by roles.<S>[hub].last_written_age_s (its journal line or session transcript; null: no sign), never by last_addressed_age_s ([pace] @hub etc.): old write, fresh address: likely dead or asleep. Infer tasks only from titles, activity, last words or the journal; compare snapshots only when a previous one exists.
  3. Answer in a few plain-text lines in the user's language, prioritizing running work and anything needing attention. Summarize large stages and offer detail on request. Include question ids when they help the owner answer. Keep raw JSON and command names out of the answer unless requested. In Claude Code terminal 2.1.287+, add: “/agent-top opens the live pane” (in the user's language). Finish after the text answer; status is read-only. Omit the pane hint in Codex, VS Code chat, claude -p and Remote Control views.

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.

Check your setup, free