Spec2Ship automates software development by orchestrating multi-agent workflows on Claude Code. It enables teams to collaborate on specs, design, and implementation. Connects to Claude Code for AI-driven development.
git clone https://github.com/spec2ship/spec2ship.gitSpec2Ship is a Claude Code plugin that adds structured orchestration to software development through multi-agent deliberation. Instead of prompting Claude Code directly, it runs roundtable discussions where 12 specialized agents—including product managers, architects, security champions, and QA leads—debate and refine decisions collaboratively. Participants respond blind to each other, preventing anchoring and surfacing genuine disagreements. The framework produces standards-based artifacts (arc42, ISO 25010, MADR) with full audit trails, making decisions auditable and reviewable. Choose Spec2Ship when decisions require multiple expert perspectives, adversarial review of security or scalability concerns, or auditable standards-based documentation.
1. **Prepare Inputs**: Gather your spec document (URL or Markdown) and define roles (e.g., @dev1, @pm). Use a GitHub/GitLab repo for CI/CD. 2. **Run Spec2Ship**: Paste the prompt template into Claude Code with your [PLACEHOLDERS] filled. 3. **Review Output**: Check the Mermaid diagram and task table for gaps. 4. **Execute**: Use the pseudocode and CI/CD config as starter templates. 5. **Iterate**: Share the output with your team in a PR or Slack thread for feedback. Tip: For large specs, break them into phases (e.g., "Phase 1: Auth") and run Spec2Ship per phase.
Requirements refinement through multi-perspective deliberation
Architecture decisions with adversarial security and feasibility review
Edge case and threat modeling across specialized agent roles
Auditable decision documentation for team handoff and compliance
No install command available. Check the GitHub repository for manual installation instructions.
git clone https://github.com/spec2ship/spec2shipCopy the install command above and run it in your terminal.
Launch Claude Code, Cursor, or your preferred AI coding agent.
Use the prompt template or examples below to test the skill.
Adapt the skill to your specific use case and workflow.
Act as a Spec2Ship orchestrator for [PROJECT_NAME]. Use the following steps to automate development: 1) Review the spec at [SPEC_URL] and identify gaps or ambiguities. 2) Generate a high-level architecture diagram in Mermaid format. 3) Decompose the spec into [NUMBER] tasks for [TEAM_MEMBER_ROLES]. 4) For each task, write pseudocode that a junior developer could implement. 5) Generate a CI/CD pipeline configuration for [DEPLOYMENT_TARGET]. 6) Summarize the plan in a Markdown table with columns: Task, Owner, Estimated Hours, Priority, and Dependencies. Respond only with executable code and structured data—no explanations.
```mermaid
graph TD
A[Spec: User Authentication] --> B[Task 1: Database Schema Design]
A --> C[Task 2: API Endpoints]
A --> D[Task 3: Frontend Components]
B --> E[Subtask: Users Table]
C --> F[Subtask: POST /login]
D --> G[Subtask: Login Form]
```
### Task Breakdown
| Task | Owner | Estimated Hours | Priority | Dependencies |
|------|-------|-----------------|----------|--------------|
| Design `users` table | @dev1 | 4 | High | None |
| Implement `POST /login` | @dev2 | 6 | High | Schema |
| Build Login Form | @dev3 | 8 | Medium | API |
### Pseudocode
**Task 1: Database Schema**
```sql
CREATE TABLE users (
id SERIAL PRIMARY KEY,
email VARCHAR(255) UNIQUE NOT NULL,
password_hash VARCHAR(255) NOT NULL,
created_at TIMESTAMP DEFAULT NOW()
);
```
**Task 2: API Endpoint**
```python
@app.post("/login")
def login(user: UserLogin):
user = db.query_user(user.email)
if not verify_password(user.password_hash, user.password):
raise HTTPException(status_code=401)
return {"token": generate_jwt(user.id)}
```
### CI/CD Pipeline (GitHub Actions)
```yaml
name: Deploy Auth Service
on: [push]
jobs:
test:
runs-on: ubuntu-latest
steps:
- uses: actions/checkout@v4
- run: pytest tests/auth_test.py
deploy:
needs: test
runs-on: ubuntu-latest
steps:
- uses: actions/checkout@v4
- run: docker build -t auth-service .
- run: docker push ghcr.io/team/auth-service:latest
```
**Risks Identified:**
- Password hashing algorithm not specified (defaulting to bcrypt).
- No rate limiting on `/login` (add in Sprint 2).
**Next Steps:**
1. @dev1 to review schema with @pm.
2. @dev2 to implement API with OpenAPI validation.
3. @dev3 to mock UI components for review.invisible human verification
A proxy. An Envoy controller. An out-of-process SDK. Power.
Create and collaborate on interactive animations with powerful, user-friendly tools.
AI assistant built for thoughtful, nuanced conversation
Enterprise workflow automation and service management platform
Automate your spreadsheet tasks with AI power
Take a free 3-minute scan and get personalized AI skill recommendations.
Take free scan