agentskills.codes
SP

speckit-specify

Create or update feature specifications from natural-language requests for this Spec Kit workflow. Use when starting a new feature or re-baselining spec.md before clarify/plan.

Install

mkdir -p .claude/skills/speckit-specify-masteryxiao && curl -L -o skill.zip "https://agentskills.codes/api/skills/download/13359" && unzip -o skill.zip -d .claude/skills/speckit-specify-masteryxiao && rm skill.zip

Installs to .claude/skills/speckit-specify-masteryxiao

Activation

This is the description your AI agent reads to decide when to run this skill — the better it matches your request, the more reliably it fires.

Create or update feature specifications from natural-language requests for this Spec Kit workflow. Use when starting a new feature or re-baselining spec.md before clarify/plan.
176 chars✓ has a “when” trigger

About this skill

Speckit Specify

Overview

Create a feature branch and a specification file, then produce a quality checklist that validates specification completeness before planning.

Workflow

  1. Parse the feature description from user input. If missing, stop with an explicit error.
  2. Generate a concise short-name for the feature branch (2-4 words, action-noun style when possible).
  3. Determine next feature number by checking all three sources:
  • Remote branches (git ls-remote --heads origin pattern match)
  • Local branches (git branch pattern match)
  • .specify/specs/ directories ([0-9]+-short-name)
  1. Run .specify/scripts/bash/create-new-feature.sh exactly once with --json, --number, and --short-name.
  2. Parse script JSON response and capture at minimum:
  • BRANCH_NAME
  • SPEC_FILE
  1. Load .specify/templates/spec-template.md and fill required sections with user-centered, testable, technology-agnostic requirements.
  2. Allow at most 3 [NEEDS CLARIFICATION: ...] markers for high-impact unknowns only.
  3. Create FEATURE_DIR/checklists/requirements.md using checklist structure from .specify/templates/checklist-template.md.
  4. Validate spec quality against checklist. Iterate updates up to 3 passes when non-clarification issues remain.
  5. If clarification markers remain, ask up to 3 structured questions, apply answers to spec, and re-run validation.
  6. Report completion with branch name, spec path, checklist outcome, and next-skill recommendation ($speckit-clarify or $speckit-plan).

Hard Rules

  • Focus on WHAT and WHY, not implementation HOW.
  • Keep all functional requirements testable and unambiguous.
  • Keep success criteria measurable and technology-agnostic.
  • Run branch/spec initialization script once per feature request.

References

  • Legacy prompt details: references/legacy_prompt.md

Search skills

Search the agent skills registry