Use when running a Spec Kit specify step to create or update a feature specification from a natural-language feature request.
Scanned 9/5/2026
Install to Claude Code
npx -y skills add me-cedric/agent-compass --skill speckit-specify --agent claude-codeInstalls into .claude/skills of the current project.
Are you the author of Speckit Specify?
Add the live security badge to your README — it updates automatically with every re-scan.
[](https://www.skillsdirectory.com/skills/me-cedric-speckit-specify)More formats (shields.io, HTML) on the badges page.
---
name: speckit-specify
description: >
Use when running a Spec Kit specify step to create or update a feature
specification from a natural-language feature request.
risk_level: medium
writes_files: true
requires_tools: []
---
# Spec Kit Specify
Read `AGENTS.md`, existing `specs/`, `.specify/README.md` if present, and the
active spec template.
## Do
1. Reuse an existing spec when the feature already exists.
2. Create one feature spec at a time under `specs/FEATURE_ID/`.
3. Write user value, actors, scenarios, requirements, non-goals, edge cases, and
measurable success criteria.
4. Keep technology choices and implementation details out of the spec.
5. Use at most three `[NEEDS CLARIFICATION]` markers, only for blocking choices.
## Done
- `spec.md` is behavior-focused and testable.
- Assumptions and open questions are explicit.
- Agent stops before plan unless user asked to continue.
Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.
No comments yet. Be the first to comment!
Adversarial multi-agent planning skill. Self-orchestrates 5 hostile category members (unspecified-low, unspecified-high, deep, ultrabrain, artistry) via team-mode for ruthless cross-critique debate, distills only the defensible insights, then MANDATORILY hands the distilled insight bundle to the `plan` agent for executable plan formalization. Use when planning needs maximum rigor and surfacing of weak assumptions, blind spots, and over-engineering. Triggers: 'hyperplan', 'hpp', '/hyperplan', ...