DAG
Provides intelligent, version-aware access to library documentation and APIs to reduce AI code hallucinations.
About
DAG (Documentation Augmented Generation) is a Model Context Protocol (MCP) server designed to combat AI code hallucinations by offering intelligent, version-aware access to software library documentation and APIs. It indexes npm library source code and documentation through a robust pipeline involving AST parsing, semantic chunking, and vector embeddings stored in Qdrant. This enables developers and AI agents to perform semantic searches across documentation and code, validate API usage in real-time against indexed signatures, and compare API changes between different library versions, ensuring accurate and up-to-date information across multiple programming ecosystems.
Key Features
- Multi-Ecosystem: Designed for extensibility to Python, Ruby, Java, etc.
- 0 GitHub stars
- Semantic Search: Vector-based search across library documentation and source code
- API Validation: Real-time validation of API usage against indexed signatures
- Hybrid Search: Combines vector similarity with keyword matching for accuracy
- Version Comparison: Diff analysis showing added/removed APIs between versions
Use Cases
- Reduce AI code hallucinations by providing accurate, version-aware library context to generative AI models.
- Enhance AI coding assistants with semantic search, API validation, and version comparison capabilities.
- Quickly validate API usage against specific library versions to ensure correctness and prevent errors.
