ibaz/paparats-indexer repository overview

⁠paparats-indexer

Automated repo indexer for Paparats MCP⁠. Clones Git repositories, indexes code for semantic search, and keeps your search index fresh on a configurable schedule.

Part of the Paparats semantic code search stack. Designed for server/production deployments where you want headless, scheduled indexing without manual CLI runs.

⁠What It Does

• Clones repos from GitHub (public or private with token)
• Indexes code using AST-aware chunking + Jina Code Embeddings via Ollama
• Runs on a schedule via cron (default: every 6 hours)
• HTTP trigger for on-demand reindexing
• Health endpoint for monitoring

⁠Quick Start

Used as part of the full Paparats server stack:

npm install -g @paparats/cli
paparats install --mode server --repos org/repo1,org/repo2

Or run directly with Docker Compose:

services:
  indexer:
    image: ibaz/paparats-indexer
    environment:
      PAPARATS_URL: http://paparats:9876
      REPOS: org/repo1,org/repo2
      CRON: "0 */6 * * *"
      # GITHUB_TOKEN: ghp_xxx  # for private repos
    volumes:
      - indexer_repos:/data/repos
    depends_on:
      - paparats
      - ollama
      - qdrant

  paparats:
    image: ibaz/paparats-server
    environment:
      OLLAMA_URL: http://ollama:11434
    ports:
      - "9876:9876"

  ollama:
    image: ibaz/paparats-ollama

  qdrant:
    image: qdrant/qdrant
    ports:
      - "6333:6333"

volumes:
  indexer_repos:

⁠Environment Variables

⁠HTTP API

Trigger reindex:

# Reindex all repos
curl -X POST http://localhost:9877/trigger

# Reindex specific repos
curl -X POST http://localhost:9877/trigger \
  -H 'Content-Type: application/json' \
  -d '{"repos": ["org/repo1"]}'

Health check:

curl http://localhost:9877/health

⁠How It Works

1. On startup and on each cron tick, the indexer iterates over configured repos
2. For each repo: clones (first run) or pulls latest changes via simple-git
3. Reads .paparats.yml from the repo root for indexing config
4. Calls the Paparats server indexing API to process files
5. Files go through AST chunking, symbol extraction, embedding, and Qdrant storage

⁠Use Cases

• Team server - shared semantic search across all team repos
• CI/CD integration - trigger reindex on push via HTTP API
• Support bots - keep code knowledge fresh for product support agents
• Multi-repo workspaces - index backend + frontend + infra repos together

⁠Tags

• latest - most recent stable build
• x.y.z - version-pinned releases (e.g., 0.2.2)

⁠Related Images

• ibaz/paparats-server⁠ - MCP server for semantic code search
• ibaz/paparats-ollama⁠ - Ollama with pre-baked code embeddings model

⁠Links

• Paparats MCP⁠ - semantic code search for AI coding assistants
• Qdrant⁠ - vector database

⁠License

MIT - github.com/bkbhub/paparats-mcp⁠