chidekina/aria-superpowers/spec
Claude Code and agent skill by chidekina/aria-superpowers
- community
- ext:[{"type":"hook","description":"Pre/post lifecycle hooks"}]
Checking that the SKILL.md file is still in the repository.
Convert PRD or requirements into structured YAML spec with effects, constraints, and outputs. Pact-lang inspired. Triggers on: create spec, spec for, convert to spec, yaml spec, structured spec.
Install
npx vskill@latest install chidekina/aria-superpowers/specSemi-extensibleMentions customization, but needs skill-specific knowledge. Read the extensible skills standard
SKILL.md
Spec Generator (Pact-lang Inspired)
Converts natural language requirements or PRDs into structured YAML specifications that are machine-readable and AI-optimized.
The Pipeline
This skill is the middle layer that adds:
- Effects declarations (what systems are touched)
- Structured outputs (all possible responses)
- Constraints (validation rules)
- Totality requirements (exhaustive error handling)
- Latency budgets (performance SLOs)
Reasoning Process
Before generating the spec YAML, reason through the PRD explicitly. Wrap your reasoning in <analysis> tags:
Only after completing the analysis, generate the YAML spec.
Output Format
Generate a YAML spec file at tasks/spec-{feature-name}.yaml:
Conversion Rules
From PRD to Spec
- Feature title →
title - Requirements →
endpointswith inputs/outputs - Data models →
typeswith constraints - "Must handle errors" →
isTotal: true - "Real-time/fast" →
latencyBudgetMs: 50-100 - "Batch/heavy" →
latencyBudgetMs: 1000-5000
Effect Detection
| Keyword in PRD | Effect | |
|---|---|---|
| "save", "create", "update", "delete" | db-write | |
| "fetch", "list", "get", "query" | db-read | |
| "notify", "send message" | whatsapp-send | |
| "create PR", "push" | github-pr | |
| "container", "docker" | docker | |
| "calendar", "schedule" | calendar | |
| "email", "send email" | email-send |
Constraint Extraction
| PRD phrase | Constraint | |
|---|---|---|
| "required" | required: true | |
| "1-200 characters" | minLen: 1, maxLen: 200 | |
| "email format" | format: email | |
| "unique" | unique: true | |
| "auto-generated" | generated: true | |
| "cannot change" | immutable: true |
Output Mapping
| Scenario | HTTP Status | Error Code | |
|---|---|---|---|
| Success (create) | 201 | - | |
| Success (read) | 200 | - | |
| Success (update) | 200 | - | |
| Success (delete) | 204 | - | |
| Not found | 404 | NOT_FOUND | |
| Invalid input | 400 | VALIDATION_ERROR | |
| Duplicate | 409 | CONFLICT | |
| Unauthorized | 401 | UNAUTHORIZED | |
| Forbidden | 403 | FORBIDDEN | |
| Rate limited | 429 | RATE_LIMITED | |
| Server error | 500 | INTERNAL_ERROR |
3-File Spec Structure (SDLD Pattern)
For larger features, generate three files instead of one — each serves a different audience and phase:
tasks/requirements-{feature}.md (EARS notation)
Machine-readable requirements using EARS (Easy Approach to Requirements Syntax):
tasks/design-{feature}.md (sequence diagrams + decisions)
Architecture decisions and interaction flows:
sequenceDiagram participant Client participant API participant DB Client->>API: POST /resource {payload} API->>API: validate(payload) API->>DB: insert(record) DB-->>API: record API-->>Client: 201 {record}
tasks/tasks-{feature}.md (execution checklist)
Ordered task list with status tracking — this replaces the need to re-read the full spec mid-sprint:
When to use 3-file vs single YAML:
- Single YAML (
spec-{feature}.yaml): small features, 1-3 stories, one developer - 3-file SDLD: larger features, 4+ stories, multi-layer changes, or when human review is important before coding starts
Automation Hooks
Auto-trigger after /prd
When a PRD is created, automatically suggest running /spec:
Integration with /ralph
The /ralph skill can read the spec file to:
- Auto-populate effects for each story
- Set isTotal flags based on spec
- Generate acceptance criteria from outputs
- Set latency budgets
Pre-commit Validation
Add to .claude/hooks/pre-commit:
Example Conversion
Input (from /prd)
Output (from /spec)
Checklist Before Saving
- [ ] All endpoints have effects declared
- [ ] All endpoints have outputs for success AND error cases
- [ ] Critical paths have
isTotal: true - [ ] Latency budgets are realistic
- [ ] Types have appropriate constraints
- [ ] Test references cover all endpoints
- [ ] Quality gates are defined
Security
- Security scanPassed, 90 of 100Pattern checks on the skill's files.
- Model reviewPassed
Quality evaluation
- Task evaluationNot runNo evaluation yet. Run one in Skill Studio.
Source
- SKILL.md in the repositoryCheckingskills/spec/SKILL.md
Popularity
- 7-day trend score
- 0
Not refreshed yet.
Extension points
hookPre/post lifecycle hooks
Works with
Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.
- Claude Code
- Cursor
- GitHub Copilot
- Windsurf
- Codex
- Gemini CLI
- Cline
- Amp
- Roo Code
- Goose
- Continue
- Kilo Code
- Junie
- Trae
- OpenHands
- Qwen Code
- AI Agent
- Tabnine Chat
- Sourcegraph Cody
- Amazon Q
- Firebase Genkit
- Sweep
- Devin
- Phind
- Mentat
- Claude Desktop
- Aider
- GPT Pilot
- Plandex
- bolt.new
- v0
- Replit Agent
- Anse
- TypingMind
- BetterChatGPT
- LibreChat
- Warp AI
- Pieces
- Fig AI
Embed badge
[](https://verified-skill.com/skills/chidekina/aria-superpowers/spec)