agentleFS
Sign inStart free
58254727-blip58254727-blip/codex-task-control-tower/skills/task-control-tower/SKILL.md

task-control-tower

Skill1 starsChanged 3 months agoMITView on GitHub

Use when coordinating or checking multiple Codex tasks and the user needs concise, evidence-backed status, stall detection, or a safe next-action board without changing those tasks.

The file

We split this skill into the rules it gives an agent

9 rules. About 0 are common, 9 are distinctive, and 0 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈9

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 0

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Treat only concrete artifacts as progress:

    line 12
  2. Distinctive:

    test result tied to the requested behavior

    line 16
  3. Distinctive:

    commit or pull request containing the work

    line 17
  4. Distinctive:

    Use exactly one state:

    line 26
  5. Distinctive:

    Read the latest compact task snapshot.

    line 37
  6. Distinctive:

    If reading fails, retry once with minimum input.

    line 39
  7. Distinctive:

    If task tools are unavailable, use the manual status template below.

    line 40
  8. Distinctive:

    After two equivalent failures, stop that path.

    line 43
  9. Distinctive:

    Keep the result compact. Report only a completion, a real blocker, a stall threshold, or a required user action.

    line 59

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “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

Task Control Tower

Inspect first. This skill is read-only unless the user separately authorizes intervention.

Evidence Standard

Treat only concrete artifacts as progress:

  • file change with a relevant diff or timestamp
  • command with meaningful output
  • test result tied to the requested behavior
  • commit or pull request containing the work
  • runtime log or process evidence
  • agent or subagent result with an inspectable artifact
  • receipt from an external operation

active is not proof of progress. idle is not proof of failure. Never infer completion from a spinner, elapsed time, or status metadata alone.

Classify Each Task

Use exactly one state:

  • completed: the requested outcome has verification evidence.
  • in_progress: recent evidence shows the task is advancing.
  • blocked: a specific dependency or authorization prevents the next safe action.
  • unverified: available evidence is too weak or cannot be retrieved.

For every task report: Status, Evidence, Evidence time, and Next step.

Procedure

  1. Read the latest compact task snapshot.
  2. Extract the newest concrete evidence and its time.
  3. If reading fails, retry once with minimum input. Do not repeat broad reads.
  4. If task tools are unavailable, use the manual status template below.
  5. At 20 minutes without new evidence, label a stall warning and name the last evidence.
  6. At 30 minutes without new evidence, recommend a safe pause that preserves current work.
  7. After two equivalent failures, stop that path. Do not make a third blind attempt.

Manual status template:

Task: <generic name>
Status: unverified
Evidence: unavailable
Evidence time: unknown
Next step: provide the latest command, test, diff, or log

Boundaries

Unless explicitly authorized, do not message another task, do not stop a process, do not modify files, and do not reprioritize work. Never expose credentials, private data, raw conversation history, or sensitive identifiers in the board.

Keep the result compact. Report only a completion, a real blocker, a stall threshold, or a required user action.

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