agentleFS
rolandtolnayrolandtolnay/llm-toolkit/skills/scope/SKILL.md

scope

Skill1 starsChanged 3 months agoMITView on GitHub

Turn a vague prompt into a well-bounded task — ask only the questions the model can't answer itself, confirm the frame, then implement. Use at the start of a fresh conversation when the request lacks bounds or success criteria.

42 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use frames, develop 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.
  • sanmidableSkill

    frames

    Also in 3 projects that use this

  • hong9LeeSkill

    develop

    빌드, 아키텍처, git 관련 작업 시 OMS MSA 공통 인프라 규칙을 적용한다. 기술 스택, 빌드 명령, 헥사고날 아키텍처 구조, git 워크플로우, AI Context 동기화 규칙을 다룬다. 코드 작성 컨벤션은 /convention, 테스트 컨벤션은 /test-guide 스킬을 참조.

    Also in 6 projects that use this

  • minhtran3124Skill

    visual

    Skill graph visualizer — scans all SKILL.md files in skills/, builds a directed relationship graph from explicit `relationships` frontmatter (falling back to content inference), and generates both a Mermaid markdown summary and an interactive HTML graph at .docs/skill-graph.html. Trigger: /visual

    Also in 3 projects that use this

  • MicrosoftSkill

    sync

    Sync the current session branch with its upstream branch, or publish the current session branch to a remote. Use when the user asks to sync a branch, pull latest changes, rebase onto upstream, push current branch, publish branch, or set upstream.

    Also in 7 projects that use this

  • nicolasestremSkill

    document

    Create focused documentation for specific components or features

    Also in 5 projects that use this

  • Skeln-ioSkill

    copy

    Direct response copywriting using proven frameworks. Brand-aware — reads ICP, voice, and tone from your brand config. Covers product descriptions, SEO meta, email, social, press angles.

    Also in 3 projects that use this

The file

We split this skill into the rules it gives an agent

3 rules. 0 are common, 3 are distinctive, and 0 are very specific to this project.

  • 0

    Common

    Most of the 39 edited versions we read keep these.

  • 3

    Distinctive

    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:

    If no prompt was given, ask once: "What do you want done, and where?"

    line 9
  2. Distinctive:

    Ask the remaining gaps in one batched round via the built-in question tool — usually 1–3 questions, each with a sensible default so the …

    line 24
  3. Distinctive:

    No question asked that the repo, a default, or a quick look could have answered

    line 46

“Common” is measured: kept by at least 6 in 10 of the 39 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

Take the user's prompt and close only the gaps that would make a frontier model guess badly. Ambiguity invites 10x the work: without bounds and a definition of done, the model fills in the blanks and over-builds. Your job is a tight frame, not a PRD — describe the destination, never the steps.

If no prompt was given, ask once: "What do you want done, and where?"

1. Check the frame

Judge the prompt against four fields. Mark each present / inferable / missing:

  1. Outcome — what the user can do or see when it works, stated as a result. This is what done looks like.
  2. Bounds — the sandbox: which area or module the work attaches to, what's off-limits, and where the relevant information lives (repo area, doc, ticket, URL) when it isn't discoverable.
  3. Verification — how to check the work: the observable signal or command that proves the outcome, and the existing behavior that must keep working.
  4. Hard constraints — only if real: a fixed API shape, required integration, specific data.

Infer over interrogate. Do a quick orientation pass (a minute, not an investigation) to resolve what the repo can answer: conventions, stack, existing patterns, likely anchor. Never ask about those. A field is a question only when the answer would materially change the implementation and can't be read from the code or reasonably defaulted.

2. Ask the gaps

Ask the remaining gaps in one batched round via the built-in question tool — usually 1–3 questions, each with a sensible default so the user can accept fast. Lead with outcome if it's vague. If nothing is missing, skip straight to step 3.

3. Confirm the frame

Restate your understanding compactly:

## Task frame
**Goal:** <outcome, as a delta from current behavior>
**Bounds:** <where the work lives; what stays untouched>
**Verification:** <how you'll prove it works; what must not break>
**Out of scope:** <what you will deliberately not do>

Then ask for approval via the built-in question tool: proceed, or adjust. Incorporate any correction and re-confirm only if the goal itself changed.

4. Implement

On approval, implement the frame end to end. Stay inside the bounds; match the conventions the orientation pass found; don't add features or abstractions beyond what the frame requires. Run the verification you named — tests, build, or driving the actual behavior — before reporting. Report faithfully: what you verified, what you couldn't, anything deliberately left out of scope.

Success criteria

  • No question asked that the repo, a default, or a quick look could have answered
  • Frame confirmed by the user before any implementation
  • Implementation stays within the confirmed bounds
  • The named verification actually ran, and the final report distinguishes verified from unverified

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.