OpenSourcePilot ✈️

The complete AI-powered open-source contribution platform. OpenSourcePilot helps developers discover the right issues, analyze codebases, plan solutions, generate test suites, and draft pull requests in one unified workflow.

---

🌐 Live Platform

• Web Client: https://open-source-pilot.vercel.app/
• Interactive API Docs: https://web-production-5369f.up.railway.app/docs
• GitHub Repository: https://github.com/coderleeon/OpenSource-Pilot

---

🧭 The Contributor Journey

OpenSourcePilot bridges the gap between searching for issues and writing pull requests. It organizes the contribution lifecycle into a logical, three-phase flow:

    ┌─────────────────────────┐      ┌─────────────────────────┐      ┌─────────────────────────┐
    │     1. DISCOVER         │ ───> │       2. ANALYZE        │ ───> │      3. CONTRIBUTE      │
    │  Open Source Radar      │      │   Repository Insights   │      │    AI-Guided Coding     │
    └─────────────────────────┘      └─────────────────────────┘      └─────────────────────────┘

1. Discover (Open Source Radar): Scan the open-source ecosystem for issues matching your skills, experience, and interests. Assess community health, merge confidence, and contributor fit before starting.
2. Analyze (Codebase Exploration): Understand the target repository's internal modules, structure tree, README conventions, and missing capability areas.
3. Contribute (Solution & PR Drafts): Generate target-relevant files lists, step-by-step solution drafts, unit/integration test suites, and review-ready pull requests.

---

⚡ Core Capabilities

📡 Phase 5: Open Source Radar

Search and evaluate GitHub repositories to find high-value open-source opportunities tailored to you.

• Dynamic Contributor Fit Scoring: Grade issue compatibility (0-100) and learning return based on user profile.
• Merge Probability Engine: Assess the likelihood of PR acceptance based on maintainer turnaround and issue history.
• Repository Health Analytics: Track community metrics, release frequency, and open issue trends.
• Missing Feature Detection: Receive LLM-generated feature suggestions to pitch to project maintainers.
• Local Saved Pipeline: Track and manage saved contribution targets locally.

🔍 Repository Intelligence & Semantic Search

Parse and index codebases to locate relevant segments without manual digging.

• Structure & Stack Analysis: Extract codebase directory structures and auto-detect primary languages and framework stacks.
• ChromaDB Vector Indexing: Create code-snippet embeddings using sentence-transformers for instant code context lookup.
• Semantic Context Retrieval: Match user query descriptions directly with relevant source files.

🛠️ Solution Planning & Auto-Testing

Go from understanding an issue to a complete contribution draft.

• Actionable Contribution Plans: Auto-generate root-cause hypotheses and sequential step-by-step implementation lists.
• Multi-Language Test Generators: Generate unit and integration test boilerplate for pytest, Jest, JUnit, Go testing, and Rust testing.
• PR Draft Synthesis: Formulate pull request summaries, Conventional Commits titles, and reviewer checklists.

---

📐 Architecture

OpenSourcePilot separates orchestration, scheduling, and AI agent execution layers for testability and scaling:

flowchart TD
    Client["Next.js React Frontend"] <-->|REST API / JSON| FastAPI["FastAPI Backend Gateway"]
    
    subgraph Services ["Application Services"]
        WorkflowService["Contribution Workflow Service"]
        RadarService["Radar Service"]
        RepoService["Repo Service"]
    end
    
    FastAPI <--> Services
    
    subgraph Agents ["AI Agent Layer"]
        WorkflowAgent["Contribution Workflow Agent"]
        RadarAgent["Radar Agent"]
        RepoAgent["Repo Agent"]
        PlanningAgent["Planning Agent"]
        TestGenAgent["Test Gen Agent"]
        PRAgent["PR Agent"]
    end
    
    WorkflowService <--> WorkflowAgent
    RadarService <--> RadarAgent
    RepoService <--> RepoAgent
    
    WorkflowAgent --> PlanningAgent & TestGenAgent & PRAgent
    
    subgraph External ["Data & Context Engines"]
        GitHub["GitHub API (PyGithub)"]
        Chroma["ChromaDB Vector Store"]
        LLM["LLM Layer (OpenRouter / Claude / OpenAI)"]
    end
    
    RepoAgent <--> GitHub
    RadarAgent <--> GitHub & LLM
    PlanningAgent & TestGenAgent & PRAgent <--> LLM
    Chroma <--> WorkflowService

Loading

---

🖼️ Application Interfaces

1. Open Source Radar (Discover)

Input developer profiles and get suited contribution listings.

2. Tailored Search & Experience Filters

Refine listings by skill keywords and contribution ease levels.

3. Contributor Fit & Health Visualizers

Understand repository health and merge probability indicators.

4. Repository Structure Analyzer

Explore directory trees, tech stacks, and missing feature recommendations.

5. AI Code Synthesis & PR Drafter

Plan changes, build integration tests, and output complete Pull Request descriptions.

---

🛠️ Tech Stack

• Frontend: Next.js (TypeScript, Tailwind CSS, Lucide icons)
• Backend: FastAPI, Uvicorn, Python 3.12+
• Semantic Search: ChromaDB, sentence-transformers (all-MiniLM-L6-v2 embeddings)
• AI Engine: OpenRouter SDK (Defaulting to Anthropic Claude 3.5 Haiku)
• GitHub Integration: PyGithub, GitPython
• Testing: Pytest

---

🚀 Quick Start

1. Clone & Set Up Directory

git clone https://github.com/coderleeon/OpenSourcePilot.git
cd OpenSourcePilot

2. Configure Environment

Create a .env file in the root directory:

OPENROUTER_API_KEY=your_openrouter_api_key
GITHUB_TOKEN=your_github_personal_access_token
LLM_PROVIDER=openrouter
OPENROUTER_MODEL=anthropic/claude-3.5-haiku

3. Run Backend (Uvicorn)

# Set up virtual environment
python -m venv .venv
.venv\Scripts\activate   # Windows
source .venv/bin/activate # macOS/Linux

# Install dependencies and start server
pip install -r requirements.txt
uvicorn app.main:app --reload

API docs will be available at: http://localhost:8000/docs

4. Run Frontend Client

cd frontend
npm install
npm run dev

Frontend interface will be available at: http://localhost:3000

5. Run Verification Tests

pytest -v

---

🔌 API Endpoints Reference

Open Source Radar (Phase 5)

• POST /api/v1/radar/discover - Scan for suitable issues based on skills, tech, interest, and experience level.
• POST /api/v1/radar/repo-health - Evaluate repository health scores, velocity, and active community trends.
• POST /api/v1/radar/missing-features - Fetch missing capability suggestions for a target repository.

Repository Indexing

• POST /api/v1/repo/analyze - Clones a repository, parses structure, and indexes files into vector storage.
• POST /api/v1/search/code - Query code snippets semantically.

Issue Analysis

• POST /api/v1/issue/list - List open issues for a repository.
• POST /api/v1/issue/analyze - Retrieve specific issue details and classify type/difficulty.
• POST /api/v1/issue/complete-workflow - Execute complete synthesis workflow (Analyze ➔ Plan ➔ Test ➔ PR).

---

🗺️ Roadmap & Release History

Phase 5 (Shipped - Current)

• Open Source Radar: Match issues dynamically based on dev profile.
• Merge Confidence Estimator: Predict PR merge probability.
• Health Audits: Analyze repo release frequency and maintainer activity.
• Missing Capability Detectors: Suggest README optimizations and missing feature hooks.
• Visual Dashboard: Redesigned Next.js visual suite with localized saved pipelines.

Phase 4 & Prior

• Semantic vector retrieval using ChromaDB.
• Dynamic test script generators for JavaScript, Python, Rust, and Go.
• Contextual Conventional Commit PR drafting.
• Dockerization and multi-cloud configurations (Railway, Vercel).

---

👤 Author & Contributors

Developed and maintained by Leeon John.

• Twitter: @LeeonJohn_
• LinkedIn: Leeon John Profile
• GitHub: @coderleeon