Open-source contributor assistant
OpenMate
An AI-assisted onboarding tool that analyzes public GitHub repositories, matches contributors with real issues, and provides repository-grounded guidance for getting started.
- Role
- Full-Stack Developer
- Project type
- Open-source contributor assistant
- Period
- Oct 2026
- Next.js 16
- TypeScript
- Gemma 3 27B
- Backboard
- Render
Contributor workflow
A contributor enters their skills, focus area, experience level, and available time, then submits a public GitHub repository. OpenMate returns a focused starting recommendation with a fit explanation, estimated scope, relevant files, and steps for investigating the issue.
The workflow continues with Ask OpenMate, a repository-grounded chat that lets contributors clarify architecture and issue context without starting a new full analysis.
Recommendation pipeline
The pipeline fetches repository metadata and open issues, builds a bounded context snapshot, and filters issues before model inference. Gemma 3 27B handles architecture analysis and contributor-to-issue matching; deterministic code owns issue identity and validates suggested paths against the repository tree.
Security and observability
Repository URLs are restricted to GitHub, ingestion is bounded, and untrusted repository content is kept separate from system instructions. The deployed application keeps provider credentials server-side and uses signed chat tokens for browser sessions.
Sentry and OpenTelemetry traces make the multi-step model workflow inspectable. Telemetry is configured to omit source code, prompts, completions, and raw Backboard thread identifiers.
Tech Stack
- Next.js 16
- React 19
- TypeScript
- Tailwind CSS
- GitHub REST API
- Gemma 3 27B
- Backboard
- Zod
- Sentry
- OpenTelemetry
- Render