agentleFS
Sign inStart free
mmaripanguecmmaripanguec/dev-sdd-kit/.claude/skills/spec-create/SKILL.md

spec-create

Skill2 starsChanged 4 months agoMITView on GitHub

Generates a layered feature spec (F1-F5) using the factory template. Use when someone asks to "create spec", "specify feature" or describes new functionality before implementing it.

5 other projects use a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

6 rules. 0 are common, 5 are distinctive, and 1 is very specific to this project.

  • 0

    Common

    Most of the 5 edited versions we read keep these.

  • 5

    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. Distinctive:

    B · NEW application: no repo exists. The spec declares it.

    line 18
  2. Distinctive:

    If you cannot classify with certainty: ASK the user.

    line 22
  3. Distinctive:

    0.3 Dependencies — ask, NEVER assume: if the stories, the as-is or the packs reveal that the application depends on another system/service:

    line 30
  4. Distinctive:

    0.4 Requirement ambiguity: if the request admits interpretations that change scope, design or certification, run /clarify BEFORE F1.

    line 40
  5. Very specific:

    Create specs/$(date +%Y-%m)-$0.md from specs/_template.md:

    line 45
  6. Distinctive:

    Do not write code. Do not modify files outside specs/ and knowledge/.

    line 58

“Common” is measured: kept by at least 6 in 10 of the 5 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

Context

  • Existing specs: !ls specs/ 2>/dev/null
  • Registered repos: !. scripts/repo-lib.sh 2>/dev/null && registry_repos | tr '\n' ' '
  • Current business rules: see knowledge/business-rules.md

STEP 0 — Requirement triage (MANDATORY before F1)

0.1 Classify against repos.yaml and the mapa-sistemas index:

  • A · EXISTING application: the requirement touches registered repos. Identify WHICH ones and record it in the spec ("Requirement type" field).
  • B · NEW application: no repo exists. The spec declares it; the stack and the architecture are decided in F5, and the task plan starts with a T0 for creation (create repo → /repo-add → scaffolding → as-is and pack at birth).
  • Mixed: a combination; apply both protocols.
  • If you cannot classify with certainty: ASK the user. Do not assume.

0.2 Load context (scenario A): for each affected repo, load its pack (<prefijo>-<repo>) and the system pack (<prefijo>-sistema).

  • Pack doesn't exist? → generate it with /repo-map (or /system-map) BEFORE F1.
  • It exists? → validate freshness: scripts/freshness.sh check <pack> and scripts/assertions.sh <sistema>. Stale → regenerate before continuing.

0.3 Dependencies — ask, NEVER assume: if the stories, the as-is or the packs reveal that the application depends on another system/service:

  • Is it registered in repos.yaml? NO → STOP AND ASK the user which is its repository (URL/path) or whether it is explicitly out of scope. With the answer: /repo-add + /repo-map for the dependency.
  • YES → validate that its pack exists and is current (as in 0.2).
  • Any INCONSISTENCY (pack contradicts code or registry, false assertion, declared dependency that does not appear in the code or vice versa): ALWAYS ASK; never normalize silently.

0.4 Requirement ambiguity: if the request admits interpretations that change scope, design or certification, run /clarify BEFORE F1; its decisions feed the spec and are recorded in its Clarifications section.

Task

Create specs/$(date +%Y-%m)-$0.md from specs/_template.md:

  1. Delegate story writing to the requirements subagent (phase 1).
  2. With the stories approved by PO/TL, delegate estimation to the estimation subagent (phase 2).
  3. Iterate refinement with /spec-review until the DoR is met, and close the phase with /consistency at a FIT FOR GATE verdict (phase 3).
  4. Delegate analysis (rules, dependencies, edge cases) to the analysis subagent (phase 4).
  5. Delegate design (contracts, C4, ADRs) to the architecture subagent (phase 5).

At each human gate: STOP, present the agent's summary and wait for explicit approval. Record in the spec who approved and when. Do not write code. Do not modify files outside specs/ and knowledge/.

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.

Check your setup, free