@paparats/cli

@paparats/cli

CLI for Paparats MCP - semantic code search across repositories with AST-based chunking, symbol graph, and vector search. Designed for AI coding assistants (Claude Code, Cursor).

Features

• AST-based code chunking via tree-sitter (10 languages) with regex fallback
• Symbol graph - cross-chunk call/reference relationships
• Vector search powered by Qdrant + llama-server (Jina Code Embeddings)
• Git metadata - commit history and ticket references per chunk
• Architectural memory - agent-maintained components / decisions (ADRs) / lessons learned with server-side similarity-gate deduplication (see main README)
• Dual MCP endpoints - coding mode and support mode with different tool sets
• Docker-based deployment - one command setup with Qdrant, embed server, and MCP server

Install

npm install -g @paparats/cli

Prerequisites

• Docker + Docker Compose - runs Qdrant, embed server, and MCP server
• Node.js >= 18

Quick Start

# 1. One-time setup: starts Docker containers, downloads embedding model (~1.6 GB)
paparats install

# 2. In your project directory
cd your-project
paparats init    # creates .paparats.yml config
paparats index   # index the codebase

# 3. Keep index in sync when files change
paparats watch

# 4. Connect your IDE (Cursor, Claude Code) to the MCP server

Install Modes

# Developer mode (default) - Docker stack + local project indexing
paparats install --mode developer

# With external Qdrant (skip Qdrant Docker container)
paparats install --qdrant-url http://your-qdrant:6333

# Server mode - full Docker stack with auto-indexer for multiple repos
paparats install --mode server --repos owner/repo1,owner/repo2

# Support mode - client-only setup, connects to existing server
paparats install --mode support --server http://your-server:9876

Commands

Command	Description
paparats install	Set up Docker containers and configure IDE
paparats init	Create .paparats.yml config in current project
paparats index	Index the codebase (or reindex after changes)
paparats watch	Watch for file changes and auto-reindex
paparats search <query>	Search indexed code from terminal
paparats doctor	Check health of all services
paparats status	Show indexing status for current project

MCP Tools

Once connected, your AI assistant gets access to these tools:

Coding mode (/mcp): search_code, get_chunk, find_usages, health_check, reindex

Support mode (/support/mcp): all coding tools plus get_chunk_meta, search_changes, explain_feature, recent_changes, impact_analysis

Configuration

Project config lives in .paparats.yml:

project: my-project
group: my-group
language: [typescript]

indexing:
  paths: [src, lib]
  exclude: [node_modules, dist, '**/*.test.ts']

chunking:
  max_lines: 60
  overlap_lines: 5

metadata:
  service: my-service
  tags: [backend, api]

Related Packages

Package	Description
@paparats/shared	Shared utilities (path validation, gitignore, excludes)
ibaz/paparats-server	MCP server Docker image
ibaz/paparats-indexer	Auto-indexer Docker image
ibaz/paparats-embed	llama.cpp llama-server + llama-swap with pre-baked embedding models

Documentation

See the full documentation for detailed setup guides, architecture overview, and configuration reference.

License

MIT