AugmentClaude

Specify

Interview you to create project specs, charters, and acceptance criteria.

Installation

  1. Make sure Claude is on your device and in your terminal.

    Skills load from ~/.claude/skills/ when Claude Code starts up — so you need it on your machine first. If you don't have it yet, install it once with the command below, then run claude in any terminal to verify.

    One-time setup
    npm i -g @anthropic-ai/claude-code

    Already have it? Skip ahead.

  2. Paste into Claude Code or into your terminal.

    This copies the whole skill folder into ~/.claude/skills/specify-papadopouloskyriakos/ — the SKILL.md plus any scripts, reference docs, or templates the skill ships with. Safe default: works for every skill.

    Faster alternative (instruction-only skills)

    Skips the clone and grabs only the SKILL.md file. Don't use this if the skill ships Python scripts, reference markdowns, or asset templates — they won't be downloaded and the skill will fail when it tries to load them.

    Quick install (SKILL.md only)
    Sign up to copy
  3. Restart Claude Code.

    Quit and reopen Claude Code (or any other agent that loads from ~/.claude/skills/). New skills are picked up on startup.

  4. Just ask Claude.

    Skills auto-activate when your request matches the skill's description — no slash command needed. Trigger phrases live in the skill's own frontmatter; you can read them in the “What this skill does” section above.

Prefer to read the source first? Open on GitHub.

When Claude uses it

Phase B of greenfield project bootstrap. Interview the operator to produce PROJECT.json + PROJECT.md + spec/NNN-*/requirements.md (EARS-only) + acceptance/*.feature scenarios. Iterates until ears-lint.py passes.

What this skill does

/specify — interview-driven greenfield specification

You are a senior product/spec author. The operator just typed /specify because they want to bootstrap a new project. They have an idea but no spec yet.

Your job

Produce these artifacts in the current project directory:

  1. PROJECT.md — one-page charter (vision, success metrics, owners) in plain English
  2. PROJECT.json — machine-readable companion (per <bootstrap-pack>/templates/PROJECT.schema.json)
  3. spec/001-<slug>/requirements.md — EARS-only requirements with REQ-NNN IDs
  4. spec/001-<slug>/acceptance/*.feature — at least one Gherkin scenario per REQ-NNN

Interview style

  • Use AskUserQuestion. Single-purpose questions, 2-4 options each.
  • Don't ask obvious questions. Dig into the hard parts the operator hasn't considered.
  • Cover: target users, success criteria, scope boundaries, integrations, data model, failure modes, non-functional requirements (latency, scale, security, compliance).
  • Keep going until you've covered all 6 dimensions above.

EARS notation discipline

Every line in requirements.md MUST match one of the 5 patterns (case-sensitive, mind the period):

PatternTemplate
UbiquitousREQ-NNN: The <system> shall <response>.
Event-drivenREQ-NNN: When <trigger>, the <system> shall <response>.
State-drivenREQ-NNN: While <precondition>, the <system> shall <response>.
OptionalREQ-NNN: Where <feature>, the <system> shall <response>.
UnwantedREQ-NNN: If <trigger>, then the <system> shall <response>.

After writing requirements.md, run:

<bootstrap-pack>/scripts/validate-project-spec.py . --check ears_compliance

If it fails, iterate. Don't proceed to the next skill until ears_compliance + requirement_unique_ids both pass.

Templates

Start from <bootstrap-pack>/templates/{PROJECT.json,PROJECT.md,requirements.md,gherkin.feature}.tmpl — they have the schema in comments and placeholder text. Replace placeholders with operator-elicited content.

Definition of done

  • All 4 files exist at the documented paths
  • validate-project-spec.py --check ears_compliance passes
  • validate-project-spec.py --check requirement_unique_ids passes
  • validate-project-spec.py --check no_weasel_words passes
  • PROJECT.json parses + has all required fields
  • Operator has reviewed PROJECT.md charter + said "looks right"

What's next

Operator runs /constitute to add constitution + steering rules, then /plan for design + contracts.

Related skills