agentleFS
OHOKAYYEAHOHOKAYYEAH/claude-skills/skills/spec/SKILL.md

spec

Skill1 starsChanged 34 days agoMITView on GitHub

Freeze the intent agreed in a discovery thread into a short FROZEN SPEC before the build: intent, success criteria that each name how they're verified, in scope vs deferred, a feature/bugfix/refactor lens, and the approved mockup. Written to the ticket (memory only without a tracker) so it survives compaction and is the baseline /work plans from and /spec-check checks against. Fire on "spec this", "freeze/lock the spec", or chained by /work or a project ship skill. NOT /interview (intent still open), NOT /spec-check (checks a build against the spec), NOT /mockup (locks the look, which this links).

168 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use backprop, ios-clean 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.
  • JuliusBrusseeSkill

    backprop

    Bug → spec protocol. When a bug is found or a test fails, trace the cause, decide whether a new §V invariant would catch recurrence, append to §B. This is the one non-obvious thing SDD does that plan-then-execute doesn't. Triggers on test failure, bug report, post-mortem, or explicit user ask.

    Also in 4 projects that use this

  • Saqlain-811Skill

    ios-clean

    Remove the DebugBridge SPM package and all #if DEBUG wiring from an iOS app. (gstack)

    Also in 5 projects that use this

  • Saqlain-811Skill

    ios-fix

    Autonomous iOS bug fixer. (gstack)

    Also in 5 projects that use this

  • bastosSkill

    ios-design-review

    Review iOS app UI/UX design quality using Apple's design principles and the Liquid Glass-era design language from WWDC25/26. Use when asked for a design review, UX critique, visual design assessment, or evaluation of an app's interface quality, design intent, alignment with the new Apple design system, or usability. Triggers: "design review", "UX review", "visual design critique", "review the UI", "does this design work", "evaluate this interface", "design feedback", "Liquid Glass review".

    Also in 5 projects that use this

  • ShreyRaviSkill

    plan-tune

    Self-tuning question sensitivity + developer psychographic for gstack (v1: observational). Review which AskUserQuestion prompts fire across gstack skills, set per-question preferences (never-ask / always-ask / ask-only-for-one-way), inspect the dual-track profile (what you declared vs what your behavior suggests), and enable/disable question tuning. Conversational interface — no CLI syntax required.

    Also in 7 projects that use this

  • timurgaleevSkill

    document-generate

    Generate missing docs for a feature, module or project as tutorials, how-tos, reference and explanation.

    Also in 6 projects that use this

  • +2 more from Saqlain-811

The file

We split this skill into the rules it gives an agent

2 rules. 0 are common, 1 is distinctive, and 1 is very specific to this project.

  • 0

    Common

    Most of the 145 edited versions we read keep these.

  • 1

    Distinctive

    This author's own choices. Worth a look.

  • 1

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Re-freezing?: If a 🔒 FROZEN SPEC already exists for this scope (step 3's lookup), start from it and change only what the thread changed.

    line 33
  2. Distinctive:

    A repo file or scratch doc: only on the user's explicit ask.

    line 67

“Common” is measured: kept by at least 6 in 10 of the 145 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

Fire only when the intent is settled — the user says so, or an orchestrator (/work, or the project's ship skill if it has one) calls this at the start of a build. Never mid-discovery: freezing a half-formed idea turns an open question into a false requirement.

Discovery is the open-ended part of the process: ideas bounced back and forth, sometimes a /mockup or /interview. This skill closes it. It distills what was agreed into one short artifact that outlives the conversation. The build plans from it, and at the end /spec-check asks whether the build matches it. A spec that exists only in the chat gets lost when the context is compacted; a frozen one doesn't.

1. Distill

Pull the settled intent from the thread, plus any ticket, prior spec, or approved /mockup design spec in play. You're capturing a conclusion, not reopening it: fill every field the conversation already answers, and leave real blanks for step 2.

Re-freezing? If a 🔒 FROZEN SPEC already exists for this scope (step 3's lookup), start from it and change only what the thread changed. The new write supersedes it.

2. Coverage pass — thin, scope-adaptive

Catch what a tired person skips; don't interrogate. Check the core plus the one lens that matches the work type. Ask the user only about gaps that would change the build, batched, once. Anything still open becomes an explicit open question in the artifact, never a silent omission.

  • Core: who it's for; success criteria; the in/out boundary (the deferred list is as valuable as the in-scope one); risks and unknowns.
  • feature → behavior, including the error / empty / loading states, plus any performance, security or data/migration impact.
  • bugfix → repro, root-cause hypothesis, regression radius, the guard test.
  • refactor → the invariant to preserve, blast radius, how equivalence is proven.

Every success criterion names its check — test, QA: <what to walk>, or inspect: <what to read>. That turns the criterion into something /spec-check can verify and the QA pass can walk. A criterion you can't name a check for is still vague; sharpen it or move it to open questions.

3. Home — ticket first

  1. The ticket. A key in the args or in the session → use it. The project has a tracker but no ticket yet → chain /ticket with the intent line to find or file one (add its file token when no build is starting now, so the ticket lands in the backlog rather than In Progress), then use the key it returns. Write the artifact as a comment (not the description, which is the ticket's own record); the newest 🔒 FROZEN SPEC comment wins. On Jira via the Atlassian MCP, for example, that's addOrEditJiraIssueComment (addCommentToJiraIssue on a v1 connection).
  2. No tracker → a project-type memory spec-<kebab-title>.md in this session's memory dir, with its MEMORY.md pointer. It's temporary: delete the file and its pointer once the work lands, or it becomes cruft.
  3. A repo file or scratch doc only on the user's explicit ask. Say once that a spec committed to the repo goes stale the moment the build diverges from it.

Lookup order for an existing frozen spec (step 1, and /spec-check's lookup by scope): the ticket's newest 🔒 FROZEN SPEC comment → a spec-* memory matching the scope → none. /work also accepts intent the user already ratified elsewhere (a handoff, the ticket's criteria) and calls this skill only when there is none.

The artifact

A distillation, not a document. Use — for a field that doesn't apply rather than padding it.

🔒 FROZEN SPEC — <title> · <feature|bugfix|refactor> · <YYYY-MM-DD>
ticket: <KEY or —> · supersedes: <date of prior freeze or —>

Intent: <one line — what changes in the world>
For: <who it's for / who's affected>

Success criteria:
  1. <observable "done looks like"> — check: <test | QA: … | inspect: …>

In scope:
  1. <precise about boundaries: which screen, which entities, the exact rule>
Deferred / out of scope:
  - <what we chose not to do — and why, if it matters>

Look: <mockup frames path (/mockup saves approved frames to a git-ignored local dir) + its
       design calls — approved | pending the caller's front stop> | —
<lens — feature: behavior + states + NFRs | bugfix: repro / cause / radius / guard test
       | refactor: invariant / radius / equivalence check>

Open questions:
  - <unknowns to settle before or while building>

4. Hand off

  • Standalone: show the artifact and where it lives, in one glanceable line on top. Then end the turn with the question the whole skill exists for: "Frozen — change anything before we build?" This is the front checkpoint where the user reviews the spec.
  • nested (an orchestrator called this): return one line — Spec: frozen at <KEY comment | memory path> — <N> criteria, <N> open questions — plus the artifact, and return to the caller. The caller owns the approval stop and folds it into one it already has, so the user isn't asked twice.

Changes after the freeze but before the build are a re-freeze (step 1). Once the build is running, a changed intent is a new freeze, not an edit to the old one. That way /spec-check always compares the build against a fixed baseline.

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.