Simili - Issue Intelligence

AI-powered GitHub issue intelligence that automatically detects duplicate issues and finds similar issues using semantic search. Supports cross-repo search and issue transfer rules.

Features

• Duplicate Detection: Automatically find similar issues when new ones are created
• Cross-Repo Search: Search for similar issues across all enabled repositories in your organization
• Issue Transfer: Automatically route issues to the correct repository based on labels, title, or content
• Smart Comments: Post helpful comments linking to related issues
• Modular Pipeline: Customize workflow steps to optimize costs and latency
• Dual Mode: Works as both a GitHub Action and CLI extension

Quick Start

1. Set up Qdrant Vector Database

Sign up for free at Qdrant Cloud or self-host.

2. Get API Keys

• Gemini API key for embeddings (free tier available)
• Optional: OpenAI API key as fallback

3. Add Secrets to Repository

Go to Settings > Secrets and variables > Actions and add:

• GEMINI_API_KEY
• QDRANT_URL
• QDRANT_API_KEY

4. Create Configuration

Create .github/simili.yaml:

qdrant:
  url: "${QDRANT_URL}"
  api_key: "${QDRANT_API_KEY}"

embedding:
  primary:
    provider: "gemini"
    model: "gemini-embedding-001"
    api_key: "${GEMINI_API_KEY}"
    dimensions: 768

defaults:
  similarity_threshold: 0.65
  max_similar_to_show: 5
  closed_issue_weight: 0.9
  comment_cooldown_hours: 1

repositories:
  - org: "your-org"
    repo: "your-repo"
    enabled: true

5. Add Workflow

Create .github/workflows/issue-intelligence.yml:

name: Issue Intelligence

on:
  issues:
    types: [opened, edited, closed, reopened, deleted]

jobs:
  process:
    runs-on: ubuntu-latest
    permissions:
      issues: write
      contents: read

    steps:
      - uses: actions/checkout@v4

      - uses: Kavirubc/gh-simili@v1
        with:
          config_path: .github/simili.yaml
        env:
          GEMINI_API_KEY: ${{ secrets.GEMINI_API_KEY }}
          QDRANT_URL: ${{ secrets.QDRANT_URL }}
          QDRANT_API_KEY: ${{ secrets.QDRANT_API_KEY }}

6. Index Existing Issues

Install the CLI and index your existing issues:

gh extension install Kavirubc/gh-simili
gh simili index --repo your-org/your-repo --config .github/simili.yaml

Custom Bot Name (Optional)

By default, comments appear from "github-actions[bot]". To have comments appear from a custom bot name like "Simili":

Create a GitHub App

1. Go to Settings > Developer settings > GitHub Apps > New GitHub App
2. Name it "Simili" (or your preferred name)
3. Set permissions:
   • Issues: Read & Write
   • Contents: Read
4. Generate a private key
5. Install the app on your repository

Use the App Token

jobs:
  process:
    runs-on: ubuntu-latest
    steps:
      - uses: actions/create-github-app-token@v1
        id: app-token
        with:
          app-id: ${{ vars.SIMILI_APP_ID }}
          private-key: ${{ secrets.SIMILI_PRIVATE_KEY }}

      - uses: Kavirubc/gh-simili@v1
        with:
          github_token: ${{ steps.app-token.outputs.token }}
        env:
          GEMINI_API_KEY: ${{ secrets.GEMINI_API_KEY }}
          QDRANT_URL: ${{ secrets.QDRANT_URL }}
          QDRANT_API_KEY: ${{ secrets.QDRANT_API_KEY }}

CLI Commands

# Index existing issues
gh simili index --repo owner/repo --config .github/simili.yaml

# Search for similar issues
gh simili search "login bug" --repo owner/repo --config .github/simili.yaml

# Sync recent updates
gh simili sync --repo owner/repo --since 24h --config .github/simili.yaml

# Validate configuration
gh simili config validate --config .github/simili.yaml

Transfer Rules

Automatically transfer issues to the correct repository:

repositories:
  - org: "myorg"
    repo: "main-issues"
    enabled: true
    transfer_rules:
      - match:
          labels: ["backend", "api"]
        target: "myorg/backend-service"
        priority: 1
      - match:
          title_contains: ["frontend", "UI"]
        target: "myorg/web-app"
        priority: 2

Rules support:

• Labels: labels: ["backend", "api"]
• Title keywords: title_contains: ["frontend", "UI"]
• Body keywords: body_contains: ["database", "SQL"]
• Author: author: "username"

Configuration Reference

Option	Description	Default
similarity_threshold	Minimum similarity score (0-1)	0.65
max_similar_to_show	Maximum similar issues to show	5
closed_issue_weight	Weight multiplier for closed issues	0.9
comment_cooldown_hours	Hours before posting another comment	1

License

MIT