agentleFS
AnthonyPoschenAnthonyPoschen/agent-skills/skills/act-within-scope/SKILL.md

act-within-scope

Skill2 starsChanged 17 days agoMITView on GitHub

Use when deciding whether to ask the user a question, make an execution choice, or continue an in-progress task. Resolve ordinary, reversible decisions that are within scope from the request and repository evidence. Ask only when intent, authority, scope, cost, or external impact is genuinely uncertain.

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 the user as the owner of product direction and authority, not as a required checkpoint for ordinary execution.

    line 12
  2. Distinctive:

    Before beginning consequential work, inspect enough context to find uncertainties that materially change what should be built, who may …

    line 18
  3. Distinctive:

    Ask those questions together when they cannot be resolved from the request, repository, or established conventions.

    line 22
  4. Distinctive:

    Do not ask about ordinary implementation choices that have a reasonable, reversible answer.

    line 24
  5. Distinctive:

    Prefer a result the user can inspect over a hypothetical question about how to produce it.

    line 40
  6. Distinctive:

    Ask before proceeding when the decision cannot be recovered from evidence and would materially change product direction, public behavior, …

    line 47
  7. Distinctive:

    Always stop for destructive or irreversible actions, security-sensitive changes, production data changes, irreversible Git operations, or …

    line 51
  8. Distinctive:

    Do not disguise a product decision as an implementation detail.

    line 55
  9. Distinctive:

    If work reveals a broader issue outside the task, finish the requested work if it remains safe and coherent.

    line 60

≈ “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

Act Within Scope

Treat the user as the owner of product direction and authority, not as a required checkpoint for ordinary execution. Make progress with the evidence available, then show the result clearly enough for the user to redirect it.

Ask Important Questions Up Front

Before beginning consequential work, inspect enough context to find uncertainties that materially change what should be built, who may authorize it, or how much work it requires.

  • Ask those questions together when they cannot be resolved from the request, repository, or established conventions.
  • Do not ask about ordinary implementation choices that have a reasonable, reversible answer.
  • Once work begins, do not pause for a choice that the resulting diff, artifact, or direct verification can make easy to review and correct.
  • Remaining phases of a sequenced workflow already in progress are ordinary execution. Finish them. Do not stop after the first phase to ask whether to continue.

Proceed With Ordinary Execution

Choose the smallest reasonable path from the request, repository evidence, and systemic conventions. Make reversible progress within scope without waiting for confirmation.

  • State a consequential assumption briefly when it affects how the user will review the result.
  • Prefer a result the user can inspect over a hypothetical question about how to produce it.
  • Report the decision, outcome, direct proof, and remaining tradeoffs when the work is complete.

Stop For Real Boundaries

Ask before proceeding when the decision cannot be recovered from evidence and would materially change product direction, public behavior, authority, scope, cost, or external impact.

Always stop for destructive or irreversible actions, security-sensitive changes, production data changes, irreversible Git operations, or sending external communications unless the user clearly authorized the specific action.

Do not disguise a product decision as an implementation detail. Do not use this skill to widen a task beyond what the user requested.

Leave The Next Step Clear

If work reveals a broader issue outside the task, finish the requested work if it remains safe and coherent. Then report the issue, evidence, and the smallest useful next action. Encode a recurring preventable failure in a durable guard when it is within scope.

Source

Adapted from Pstack's Never Block On The Human principle.

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.