simili-web

Simili Web UI

A simple web interface for analyzing GitHub issues against the Simili Bot vector database.

Overview

This web application provides a user-friendly interface to test issue analysis without making any GitHub writes. It processes issues through the same pipeline as the bot (similarity search, duplicate detection, quality assessment, and triage) but always runs in dry-run mode.

Features

• Issue Analysis: Submit issue title and body to analyze against indexed issues
• Similar Issues: Find semantically similar issues with similarity scores
• Duplicate Detection: Check if the issue is a duplicate of existing issues
• Quality Assessment: Get quality scores and improvement suggestions
• Label Suggestions: Receive automated label recommendations
• Transfer Recommendations: See if the issue should be transferred to another repository
• Minimal UI: Clean, shadcn-inspired design with black/white theme and green/red/yellow accents

Requirements

• Go 1.21+
• Environment Variables:
  • GEMINI_API_KEY: Optional AI key (takes precedence when both provider keys are set)
  • OPENAI_API_KEY: Optional AI key (used when Gemini key is not set)
  • QDRANT_URL: Vector database URL (e.g., https://xxx.qdrant.io:6334)
  • QDRANT_API_KEY: Qdrant authentication key
  • QDRANT_COLLECTION: Collection name (e.g., ballerina-issues)
  • GITHUB_TOKEN: Optional, for fetching additional issue data

Quick Start

1. Build the application

cd cmd/simili-web
go build -o simili-web .

2. Set environment variables

export GEMINI_API_KEY="your-gemini-api-key"
# or export OPENAI_API_KEY="your-openai-api-key"
export QDRANT_URL="https://your-qdrant-instance.qdrant.io:6334"
export QDRANT_API_KEY="your-qdrant-api-key"
export QDRANT_COLLECTION="ballerina-issues"

3. Run the server

./simili-web

The server will start on http://localhost:8080 by default.

4. Change port (optional)

PORT=3000 ./simili-web

API Endpoints

GET /

Serves the static web UI (HTML, CSS, JS)

GET /api/health

Health check endpoint

Response:

{
  "status": "ok"
}

POST /api/analyze

Analyze an issue against the vector database

Request Body:

{
  "title": "Bug: HTTP connector timeout issue",
  "body": "The HTTP connector times out after 30 seconds",
  "org": "ballerina-platform",
  "repo": "ballerina-library",
  "labels": ["bug", "module/http"]
}

Response:

{
  "success": true,
  "similar_issues": [
    {
      "Number": 8634,
      "Title": "HTTP service hangs for request",
      "URL": "https://github.com/...",
      "Similarity": 0.85,
      "State": "open"
    }
  ],
  "is_duplicate": false,
  "duplicate_of": 0,
  "duplicate_reason": "Issues describe different problems",
  "quality_score": 0.85,
  "quality_issues": ["Missing error logs"],
  "suggested_labels": ["bug", "module/http", "Priority/High"],
  "transfer_target": "",
  "transfer_reason": ""
}

Architecture

Pipeline Steps

The web UI runs the following pipeline steps in order:

1. Gatekeeper: Validates the issue format and content
2. Similarity Search: Finds similar issues using vector embeddings
3. Duplicate Detector: Uses LLM to determine if issue is a duplicate
4. Quality Checker: Assesses issue quality and provides feedback
5. Triage: Suggests labels and transfer recommendations

Note: The indexer and action_executor steps are excluded to prevent any writes to the vector database or GitHub.

Dry-Run Mode

All operations run with DryRun=true, ensuring:

• No comments posted to GitHub
• No labels applied
• No issue transfers
• No vector database writes

Static Files

Static files (HTML, CSS, JS) are embedded into the binary using Go's embed package, making the application a single executable with no external dependencies.

Configuration

The web UI respects the same .simili.yaml configuration file as the CLI. You can override settings via environment variables or by modifying the config file.

Example .simili.yaml:

qdrant:
  url: https://your-instance.qdrant.io:6334
  api_key: ${QDRANT_API_KEY}
  collection: ballerina-issues

embedding:
  provider: gemini # or openai
  api_key: ${GEMINI_API_KEY} # or ${OPENAI_API_KEY}
  model: text-embedding-004 # or text-embedding-3-small
  dimensions: 768 # use 1536 for text-embedding-3-small

defaults:
  similarity_threshold: 0.75
  max_similar_to_show: 5

Development

Project Structure

cmd/simili-web/
├── main.go              # Server and API handlers
├── static/
│   ├── index.html       # Web UI structure
│   ├── styles.css       # Minimal shadcn-style CSS
│   └── app.js           # Frontend logic and API calls
└── README.md            # This file

Adding New Features

1. Backend: Modify main.go to add new API endpoints or change pipeline steps
2. Frontend: Update index.html, styles.css, or app.js for UI changes
3. Rebuild: Run go build to embed updated static files into the binary

Testing

Test the API directly with curl:

curl -X POST http://localhost:8080/api/analyze \
  -H "Content-Type: application/json" \
  -d '{
    "title": "Bug: Test issue",
    "body": "This is a test",
    "org": "ballerina-platform",
    "repo": "ballerina-library"
  }'

Troubleshooting

"No AI API key found"

Set either GEMINI_API_KEY or OPENAI_API_KEY before running the server.

"Failed to connect to Qdrant"

• Check QDRANT_URL format (include https:// and port :6334)
• Verify QDRANT_API_KEY is correct
• Ensure Qdrant instance is accessible

No similar issues found

• Verify the collection name matches your indexed data
• Check that issues have been indexed using simili index

Port already in use

Change the port: PORT=3001 ./simili-web

License

Part of the Simili Bot project.