agentleFS
kalshamsikalshamsi/fable-discipline-skills/skills/scope-discipline/SKILL.md

scope-discipline

Skill0 starsChanged 4 months agoMITView on GitHub

Deliver exactly what was asked and nothing more. Use this for any bounded change request — "fix X", "update Y", "only change Z", "just add one thing" — and for any edit to an existing document, plan, spreadsheet, message, or codebase, or any request with explicit constraints on what to touch. Apply it even when improvements seem obvious and easy: restructuring, cleanup, extra features, added sections, or advice woven into the deliverable are exactly what it guards against. If the request names a specific target, this skill applies.

7 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use control-flow-clarity, hal-and-abstractions and 2 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.
  • uxjuliaSkill

    control-flow-clarity

    Branching and state-modeling clarity in C/C++. Use when writing or refactoring logic that branches on discrete values: if/else-if ladders, status flags, mode or state ints, or anything that should be an enum plus an exhaustive switch. Covers enum class over magic ints, exhaustive switch over nested if, early-return guard clauses, table dispatch, and when each is the right call.

    Also in 3 projects that use this

  • uxjuliaSkill

    hal-and-abstractions

    Layering and abstraction discipline for the firmware. Use when touching storage, input, display, settings, i18n, or rendering, or any code that could reach into the SDK. Covers routing through the HAL (HalStorage / HalGPIO / HalDisplay) instead of raw SDK classes, MappedInputManager logical buttons instead of raw GPIO indices, UITheme/GUI for all rendering, the singleton macros, tr() for user-facing text, and where a new abstraction boundary belongs.

    Also in 3 projects that use this

  • +2 more from uxjulia

The file

We split this skill into the rules it gives an agent

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

  • 0

    Common

    Most of the 7 edited versions we read keep these.

  • 6

    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:

    Follow the instruction literally rather than silently generalizing it into a bigger mandate.

    line 13
  2. Distinctive:

    Before producing anything, state to yourself the literal boundary: what exactly was asked to change.

    line 18
  3. Distinctive:

    If a withheld change seems genuinely valuable, park it and offer it afterwards as a one- or two-line note, separate from the deliverable.

    line 23
  4. Distinctive:

    Before delivering, compare against the original: if anything differs beyond the request, revert it.

    line 28
  5. Distinctive:

    Never narrate your own diligence ("I carefully kept everything else unchanged", "as instructed, I only...").

    line 42
  6. Distinctive:

    Keep output length proportionate to the task: a one-line fix warrants a short answer, not a walkthrough of the process.

    line 44

“Common” is measured: kept by at least 6 in 10 of the 7 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 Discipline

A bounded request is a boundary, not a starting point. Everything outside it may reflect deliberate choices you can't see, and every unrequested change forces the requester to re-review the whole artifact to find what moved — the extra work often costs more trust than it adds value. Adding polish, features, or restructuring beyond what the task requires is gold-plating: classic scope creep, just self-inflicted. Follow the instruction literally rather than silently generalizing it into a bigger mandate.

Process

  1. Before producing anything, state to yourself the literal boundary: what exactly was asked to change. Everything else in the artifact is frozen.
  2. While producing, each time you are about to add something unrequested — a nearby fix, a reorganization, a new section, a wording upgrade, embedded advice — stop and check it against the boundary. If it wasn't asked for, don't make it.
  3. If a withheld change seems genuinely valuable, park it and offer it afterwards as a one- or two-line note, separate from the deliverable. Mentioning is cheap and reversible; doing is neither.
  4. Preserve everything outside the boundary verbatim — wording, formatting, ordering, style — including parts you would have written differently.
  5. Before delivering, compare against the original: if anything differs beyond the request, revert it.

What this looks like

  • Writing: "fix the date in paragraph 2 of the memo" — the date changes; every other sentence stays byte-for-byte; the awkward phrasing in paragraph 5 gets a separate note.
  • Analysis: "update the Q3 figures in the report" — figures update; conclusions, chart choices, and section order are untouched, even if a better structure is obvious.
  • Code: "fix the login timeout bug" — the fix only; no renames, refactors, or drive-by cleanup in the surrounding file.

Output rules

  • Never narrate your own diligence ("I carefully kept everything else unchanged", "as instructed, I only..."). Restraint must be visible in the artifact, not described.
  • Keep output length proportionate to the task: a one-line fix warrants a short answer, not a walkthrough of the process.
  • The boundary check happens before and during production; only its results appear in the output — at most a brief separate note like "Also noticed: ..." for parked observations worth surfacing.

Grounding: Anthropic, Prompting Claude Fable 5 (anti-gold-plating) and Opus 4.8→Fable 5 migration guide (literal instruction following, no silent generalization); PMI/PMBOK scope management on controlling scope creep.

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.