GitHub Issue Triager
https://github.com/coleam00/githubissuetriager- Category
- Developer Tools
- Rank
- No. 1575Tools index
- Pricing
- Open Source
- Type
- TOOL
- Builder
- coleam00
- GitHub
- 32 stars
- Date
About
AI-powered GitHub issue triage dashboard on Next.js 15, Neon Postgres, and pgvector that classifies, embeds, and plans implementation for every issue in a repo.
What it does
Operators sync repository issues into a dashboard, filter them by state and generated triage, inspect related issues, request an implementation plan, and record a proposed dispatch branch. Each classification remains in history while the interface presents the latest result.
Why it's ranked here
This is strongest as an inspectable reference project, not a finished automation service. It demonstrates a complete issue workflow plus unusually practical isolation patterns for parallel coding sessions. Deterministic offline substitutes keep the workflow runnable, although their output is intentionally weaker than the configured AI path.
What's good
Classification produces constrained categories, priorities, complexity, summaries, and reasoning. Historical classifications remain append-only. Similarity search uses normalized vectors and an HNSW cosine index. Separate database branches protect concurrent worktrees from each other, while path-derived ports avoid shared allocation state. Type checking and a smoke script share one validation command.
Tradeoffs
A database connection is mandatory even when AI access is absent. Offline embeddings are stable token hashes, not semantic representations, and offline plans are generic templates requiring review. Sync depends on the GitHub CLI and retrieves at most 50 issues per request. Port hashing can collide. Worktree teardown is manual, and dispatch only records a run with a generated branch name rather than starting an agent.
How to use it well
Use it as a local triage console or as source material for teams experimenting with isolated, parallel coding sessions. Configure AI access when classification quality, semantic similarity, and issue-specific plans matter. Keep human review before implementation. Pair it with separate execution and lifecycle monitoring, because its dispatch step does not run work or track real agent progress.
Technical notes+
package.json defines a private pnpm application with Next.js 15, React 19, TypeScript, raw postgres, Zod, and the OpenAI SDK. src/lib/ai.ts uses gpt-5.4-nano for classification and plans, text-embedding-3-small for 1,536-dimensional embeddings, and deterministic fallbacks without OPENAI_API_KEY. src/lib/db.ts requires DATABASE_URL during module loading and configures SSL. src/lib/github.ts shells out to gh issue list. src/app/api/sync/route.ts caps sync at 50 issues and supplies a default repository when GITHUB_REPO is missing. src/components/IssueActions.tsx sends the database issue ID to action routes, and src/app/api/plan/[id]/route.ts queries issues.id; this conflicts with README.md, which says route IDs are GitHub issue numbers. src/app/issues/[number]/page.tsx uses GitHub issue numbers for detail-page routing.
Observed
- License
- MIT
- Primary language
- TypeScript
- Packaging
- Private pnpm application installed with pnpm install
- Interfaces
- Next.js web dashboard and POST API routes
- Data platform
- Neon Postgres with pgvector and an HNSW cosine index
- Worktree platforms
- PowerShell wrapper for Windows and Git Bash shell tooling; WSL execution is rejected
- Validation surface
- TypeScript checking plus a smoke script through pnpm validate
Read from README.md, package.json, src/lib/ai.ts, src/lib/db.ts, src/app/page.tsx, src/lib/types.ts, src/lib/github.ts, src/app/layout.tsx, src/components/SyncButton.tsx, src/components/ThemeToggle.tsx, src/components/IssueActions.tsx, src/app/issues/page.tsx, src/app/api/sync/route.ts, src/app/issues/[number]/page.tsx, src/app/api/plan/[id]/route.ts.
What it can do
Classify GitHub issues into categories
GitHub repository issues → Categorized issue classifications
Generate vector embeddings for issues
GitHub issue text content → Vector embeddings stored in database
Create implementation plans for issues
GitHub issue details → Step-by-step implementation plans
Analyze repository issues automatically
GitHub repository URL or connection → Processed issue analysis results
Display issue triage dashboard
Repository issue data → Interactive dashboard interface
Store issue analysis in database
Processed issue data and embeddings → Persistent database records
Tags
Tech Stack
Comments (0)
No comments yet
Editorially curated, with community endorsements as a secondary signal. Corrections welcome.