agentleFS
gleicongleicon/devskills/skills/ds-spec/SKILL.md

ds-spec

Skill20 starsChanged 46 days agoMITView on GitHub

Convert a description into a structured specification.

1 other project uses a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

8 rules. About 0 are common, 5 are distinctive, and 3 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈5

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 3

    Very specific

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

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

    When invoked, gather information and produce a spec document that defines WHAT the system does, not HOW it is implemented.

    line 7
  2. Very specific:

    Before writing, look for RETRO.md (current directory, then repo root).

    line 9
  3. Distinctive:

    If the user has provided a description: proceed directly.

    line 13
  4. Distinctive:

    If not: ask the questions below one per message, then produce the spec without further prompting.

    line 14
  5. Distinctive:

    Every requirement is verifiable.

    line 71
  6. Distinctive:

    No ambiguous quantifiers: not "fast" or "reasonable" — state the number.

    line 73
  7. Distinctive:

    If a language profile is active, add a "Technical Profile" section:

    line 78
  8. Very specific:

    Ask: "Should I proceed to planning with /ds-roadmap?" after displaying the spec.

    line 88

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “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

When invoked, gather information and produce a spec document that defines WHAT the system does, not HOW it is implemented. The spec becomes the input to planning — hand it to /ds-roadmap to sequence the work.

Before writing, look for RETRO.md (current directory, then repo root). Its rule bullets are priors distilled from past releases by /ds-retro: cite a rule when it bears on a spec decision — the user can override it; never let it steer silently.

Process

If the user has provided a description: proceed directly. If not: ask the questions below one per message, then produce the spec without further prompting.

Questions (ask only what is missing):

  1. What is the primary user action this system enables?
  2. What does success look like — how do you know it works?
  3. What are the hard constraints: scale, latency, cost, compliance?

Spec Structure

Produce a document with these sections:

# Specification: <name>

## Problem
One paragraph. What problem does this solve and for whom?

## Scope
What is in scope. What is explicitly out of scope.

## Users
Who uses this. What they need to accomplish. No personas — just roles and goals.

## Functional Requirements
Numbered list. Each requirement is a verifiable statement.
FR-1: The system shall <verb> <object> when <condition>.

## Non-Functional Requirements
NFR-1: Latency: <P99 target> under <load description>
NFR-2: Scale: <concurrent users or requests/sec>
NFR-3: Availability: <uptime target>
NFR-4: Data retention: <policy>

## Interfaces
- API endpoints or CLI commands (names and purpose, not full schema)
- External systems this integrates with
- Data formats at boundaries

## Constraints
- Language/runtime (from language profile if set)
- Infrastructure (from project context if set)
- Forbidden approaches

## Acceptance Criteria
Numbered list. Each item is a pass/fail test that can be run.
AC-1: Given <state>, when <action>, then <outcome>.

Coverage — map every FR to the AC(s) that verify it. An uncovered FR is a spec defect:
FR-1 → AC-1, AC-2
FR-2 → AC-3

## Open Questions
Questions that must be answered before implementation begins.

Requirement Quality Rules

  • Every requirement is verifiable. If it can't be tested pass/fail, it's a wish, not a requirement — rewrite it or move it to Constraints.
  • SHALL = mandatory, SHOULD = recommended, MAY = optional. Choose deliberately.
  • No ambiguous quantifiers: not "fast" or "reasonable" — state the number.
  • One requirement per statement. Split compound requirements; list exceptions as sub-items (FR-N.a, FR-N.b).

Language Profile Integration

If a language profile is active, add a "Technical Profile" section:

  • Primary language and version
  • Runtime target
  • Build toolchain
  • Testing framework

Output

Write the spec to SPEC.md in the current directory — or to a path the user names — then display it inline for review. Never into .project/, which holds session state with one writer per file; a spec is a work product, something to read, hand to /ds-roadmap, and retire once its work has shipped.

Ask: "Should I proceed to planning with /ds-roadmap?" after displaying the spec.

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.