A Local-First Document Retrieval MCP
Mobile-Muffin2532 · reddit · 2026-07-18
nowdocs is a local-first document retrieval MCP service designed for coding agents.
The author's core motivation: developers often want agents to fetch the latest framework documentation without sending every query, embedding, and indexing task to a cloud RAG service. To solve this, nowdocs handles document ingestion, vectorization, and hybrid search entirely locally, exposing only read-only search/list tools via standard MCP input/output.
Its boundary design is also very clear:
- The MCP only retrieves cleaned documents and cannot modify local state.
- Installation, indexing, updating, and rebuilding remain explicit CLI operations with approval gates.
- Optional remote reranking is disabled by default, with a local fallback always available.
The author primarily wants feedback: for those building coding agents, is this kind of document retrieval better suited as an MCP server, a local CLI, or some other form?
More from coding & agent
- A roundup of AI agents and MCP resources, including how to evaluate agents — _jaydeepkarale · 2026-07-21
- Anthropic shares a masterclass on how it builds AI agents — _jaydeepkarale · 2026-07-21
- A beginner guide to AI agents points readers to a Stanford webinar — _jaydeepkarale · 2026-07-21
- A full course shows how to build and deploy an AI agent with OpenAI and LangChain — _jaydeepkarale · 2026-07-21
- A practical guide on how to evaluate AI agents — _jaydeepkarale · 2026-07-21
- Karpathy’s LLM fundamentals are recommended before building AI agents — _jaydeepkarale · 2026-07-21