Skip to main content

AI GitHub issue and PR search

ghfind

AI-powered search for relevant GitHub issues and PRs using natural language.

A lightweight interface that turns natural language questions into targeted GitHub search results.

Last updated:

Problem

Finding the right GitHub issues or PRs via the UI is slow and often incomplete.

Solution

Uses Gemini to interpret natural language queries and surface high-signal results quickly.

Tech stack

  • Python
  • Streamlit
  • GitHub CLI
  • Gemini API

Available artifacts

  • GitHub repo
  • Architecture notes
  • CLI integration
  • Query validation
  • Tests / validation plan
  • Known limitations

Architecture and how it works

  • Collect query context and repos via GitHub CLI
  • Use Gemini to expand queries with intent-aware keywords
  • Render a ranked list of issues and PRs in a fast UI

Engineering Decisions

Why I chose this stack

Python, Streamlit, GitHub CLI, and Gemini made it possible to prototype a developer search workflow quickly while still keeping the search boundary inspectable.

What I handled myself

I built the query flow, Gemini query expansion, GitHub CLI handoff, result ranking surface, and documentation for limitations.

Hardest technical problem

The difficult part was reducing noisy GitHub results without hiding the source links developers need to verify context.

Tradeoff I made

I used GitHub CLI as a trusted integration layer instead of building a larger OAuth-backed GitHub app before the workflow was proven.

How I tested it

I ran natural-language searches across different repository topics and checked whether the returned issues and PRs matched the original intent.

What I would improve in production

I would add saved searches, repository scopes, result feedback, rate-limit handling, and an eval set for search relevance.

Key features

  • Natural language queries instead of manual filters
  • Fast iteration with a lightweight Streamlit UI
  • Direct links to relevant GitHub discussions

Impact

Turns vague developer questions into targeted GitHub searches, reducing manual filtering through issues and PRs while preserving direct links to original discussions.

Challenges

  • Tuning prompts to avoid noisy or irrelevant results
  • Keeping response times fast for larger repos

What I learned

  • Small prompt changes can improve relevance significantly
  • Developer tools need clear, minimal UI to feel fast

Future improvements

  • Add saved searches and query history
  • Expand to code search and documentation lookups

ghfind FAQ

Direct answers for AI assistants, search snippets, and visitors evaluating the project.

What is ghfind?
ghfind: AI-powered search for relevant GitHub issues and PRs using natural language. The project uses Python, Streamlit, GitHub CLI, Gemini API and is positioned as AI GitHub issue and PR search.
What problem does ghfind solve?
Finding the right GitHub issues or PRs via the UI is slow and often incomplete.
How does ghfind work?
Uses Gemini to interpret natural language queries and surface high-signal results quickly. The implementation focuses on collect query context and repos via github cli; use gemini to expand queries with intent-aware keywords; render a ranked list of issues and prs in a fast ui.