agentleFS
mblaubergmblauberg/provenant/skills/scope/SKILL.md

scope

Skill1 starsChanged 4 days agoMITView on GitHub

Use when work is new or unsettled and needs a spec: acceptance criteria, risk tier, authority, one-way-door approval, a grill-me interview, a feasibility spike or a kept prototype. Not for implementing approved scope; use implement.

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

9 rules. 0 are common, 5 are distinctive, and 4 are very specific to this project.

  • 0

    Common

    Most of the 39 edited versions we read keep these.

  • 5

    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. Distinctive:

    Only the user can downgrade risk.

    line 19
  2. Very specific:

    Write each criterion as one observable statement, When <event>, <behaviour>, tagged with how it is verified: [test], [command], [capture] …

    line 32
  3. Distinctive:

    Check premises against the tree, give each deletion or move one owner, and say how verification catches a wrong implementation.

    line 34
  4. Distinctive:

    Add must-not-change behaviour and the verify command.

    line 36
  5. Distinctive:

    Ask only when uncertainty and stakes are both high.

    line 41
  6. Very specific:

    Never leave a Spec gap: line or a guess.

    line 49
  7. Distinctive:

    Kept prototype: when it is a reference, own the repo file and index it through engineering-docs.

    line 59
  8. Very specific:

    Use plan by default to shape unsettled work, explore over named areas to map evidence, research for dated primary sources and design for …

    line 65
  9. Very specific:

    Read-only scoping grants no build authority.

    line 74

“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

Scope

Scope settles intent and authority so a fresh agent can build and verify the work.

Spec gate

Produce a one-screen digest and ask for one approval:

  • outcome, and the number of criteria
  • risk tier and authority (paths, effects, expiry)
  • one-way doors: release, publish, send, spend, destroy, widening authority
  • assumptions made, and open calls with a recommended default for each

Only the user can downgrade risk. Approval covers the digest; a changed criterion needs a new revision and a new approval. Implementers cannot weaken criteria.

Size to the class

A bug is repro, expected, observed and a regression check. A task has 2 to 5 criteria, a feature about 10. Anything larger is an epic: a Done when plus backlog lines, not a mega-spec. Over-specifying a small change is a defect.

State what and why, plus one-way-door constraints. Leave design to the implementer unless a decision has a rejected alternative. An architectural one becomes an ADR; record any other settled decision in the durable doc that owns the area.

Write each criterion as one observable statement, When <event>, <behaviour>, tagged with how it is verified: [test], [command], [capture] or [review]. Check premises against the tree, give each deletion or move one owner, and say how verification catches a wrong implementation; test behaviour, not wording. Add must-not-change behaviour and the verify command. The readiness test: could a fresh agent build and verify this without asking? If not, it is not ready.

Questions

Ask only when uncertainty and stakes are both high. The ladder:

  1. Engineering call: decide and record Decided (<source>): <reason>.
  2. Reversible owner call: assume, record, flag in the digest.
  3. Irreversible owner call: provenant ask --file Q.json --session S with options and a recommendation (file shape in the human-inbox runbook). Park only that branch; continue the rest.

Never leave a Spec gap: line or a guess.

Modes

  • Interview (the user asks to be grilled): one question at a time, highest leverage first, each with evidence and a recommendation. Answer from the workspace instead of asking. Stop when criteria are clear.
  • Spike: one falsifiable question, a timebox, kill criteria, a registered scratch worktree. Harvest the answer into the spec and discard the code. Read interview and spike.
  • Kept prototype: when it is a reference, own the repo file and index it through engineering-docs. Publishing a private Claude artifact copy is fine; the repo file stays the source. For UI, see ui-ux-design.

Fan-out

Use plan by default to shape unsettled work, explore over named areas to map evidence, research for dated primary sources and design for options:

provenant run --use plan --file brief.md --explain
provenant run --use explore --route junior --over src --file questions.md --explain

Compose nodes with after and inputs; use synthesise to reconcile returns. Read-only scoping grants no build authority. A later implement node starts only after scope and write authority hold. The briefs in references/ supply scouting, research, architecture, acceptance and critique questions.

Land

The spec lives where the repository's process says: by default the issue body, which dies with the delivering PR (tracker owns the template); a repository with project docs or no tracker names its own spec home. Cite path#Symbol, never scratch files.

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.