mycrew: AI-Powered Development & Review Pipelines

mycrew provides two AI-powered pipelines:

• Development Pipeline: Transforms GitHub/GitLab issues into implemented code
• Review Pipeline: Comprehensive PR review with 10 parallel specialized agents

---

Quick Start

# 1. Install
pip install uv
git clone https://github.com/iklobato/mycrew.git
cd mycrew
uv sync

# 2. Configure
cp .env.example .env
# Edit .env and set OPENROUTER_API_KEY

# 3. Run
./cli.py development "https://github.com/owner/repo/issues/123"
./cli.py review "https://github.com/owner/repo/pull/123"

---

CLI Usage

# Development pipeline
./cli.py development <issue-url>
./cli.py dev <issue-url>

# Review pipeline
./cli.py review <pr-url>
./cli.py rev <pr-url>

# Options
--repo-path /path/to/repo    # Local repository path
-v                              # Verbose output

# Help
./cli.py --help
./cli.py development --help
./cli.py review --help

---

Development Pipeline

Transforms GitHub/GitLab issues into implemented code.

Issue Analyst → Explorer → Clarify → Architect → Implementer → Test Validator → Reviewer → Commit

Agent	Description
Issue Analyst	Parse issue into requirements
Explorer	Deep codebase analysis with file reading
Clarify	Identify ambiguities
Architect	Create file-level implementation plan
Implementer	Write code following architect's plan
Test Validator	Write tests covering acceptance criteria
Reviewer	Security, performance, code quality checks
Commit	Create branch, commit, PR

---

Review Pipeline

Comprehensive PR review with 10 parallel agents + Signoff.

┌─────────────────────────────────────────────────────────────┐
│              10 Parallel Agents (async_execution=True)    │
│                                                             │
│  Context | Architecture | Correctness | Security | Perf    │
│  Test Coverage | Readability | Consistency | Err Handling  │
│  Documentation                                           │
└─────────────────────────────────────────────────────────────┘
                           ↓
                    ┌─────────────┐
                    │  Signoff    │
                    └─────────────┘

Agent	Focus
Context	PR description clarity, linked issues, scope
Architecture	Design decisions, patterns, coupling
Correctness	Logic errors, edge cases, async handling
Security	Injection risks, secrets, auth/authz
Performance	N+1 queries, blocking calls, memory
Test Coverage	Unit tests, edge cases, test quality
Readability	Naming, complexity, duplication
Consistency	Style guide, naming conventions
Error Handling	Error catching, logging, retries
Documentation	Comments, API docs, changelog
Signoff	Final synthesis and recommendation

---

Environment Variables

Variable	Required	Description
OPENROUTER_API_KEY	Yes	LLM API key from openrouter.ai
GITHUB_TOKEN	No	GitHub token for issues/PRs
GITLAB_TOKEN	No	GitLab token for issues/MRs
SERPER_API_KEY	No	Web search
PROVIDER_TYPE	No	"openrouter" (default) or "huggingface"

---

Project Structure

mycrew/
├── cli.py                    # Unified CLI
├── agents/
│   ├── development/          # Development agents
│   └── review/              # Review agents (10 + Signoff)
├── pipelines/
│   ├── development/
│   └── review/
├── shared/                  # Issues, pulls, LLM, settings, tools
└── ...

---

License

MIT