Do you maintain this server? Claim this listing to add a logo or screenshot and to correct or extend it. Free, and a person checks every claim.
What it does
Easy-to-setup local RAG server with minimal configuration
Description as published by the publisher in the official MCP Registry, read 10 Oct 2026. Not written or verified by Few Servers.
Facts
| Registry name | io.github.shinpr/mcp-local-rag |
|---|---|
| Publisher | GitHub account shinpr Namespace verified by the registry through the publisher's GitHub account. |
| Category | AI, memory and agents sorted by our keyword rules |
| Transport | local package (npm) |
| Version | 0.21.1 |
| This version published | 7 Oct 2026 |
| GitHub stars | 414 on shinpr/mcp-local-rag read from GitHub on 12 Oct 2026. GitHub's count, not a rating by us. |
| Last seen | 10 Oct 2026, in our crawl of the official MCP Registry |
| Tested by Few Servers | No |
Install
- npm package, as published:
mcp-local-rag· version 0.21.3 · transport stdio
Standard run command for npm packages. The registry publishes the package name, not a command; check the repository for required arguments.npx -y mcp-local-rag
Install details read from the registry record on 12 Oct 2026. We have not run this server.
Keys and settings
The registry record declares settings, none marked secret. Declared by the publisher:
BASE_DIR(environment variable, not marked required): Base directory for document storage (defaults to current working directory). Ignored when BASE_DIRS is set.BASE_DIRS(environment variable, not marked required): JSON array of base directories (e.g. '["/a","/b"]'). Takes precedence over BASE_DIR.DB_PATH(environment variable, not marked required): Path to LanceDB database directory (defaults to ./lancedb/)CACHE_DIR(environment variable, not marked required): Directory where Transformers.js models are cached (defaults to ./models/)HF_ENDPOINT(environment variable, not marked required): Hugging Face model download endpoint. Set this to a mirror URL when direct downloads are blocked (defaults to https://huggingface.co).MODEL_NAME(environment variable, not marked required): Embedding model name (defaults to Xenova/all-MiniLM-L6-v2)MAX_FILE_SIZE(environment variable, not marked required): Maximum file size in bytes (defaults to 104857600 / 100MB)RAG_MAX_DISTANCE(environment variable, not marked required): Maximum distance threshold for filtering search results. Results with distance greater than this value will be excluded. Lower values mean stricter filtering (eRAG_GROUPING(environment variable, not marked required): Grouping mode for quality filtering. 'similar' returns only the most similar group (stops at first distance jump). 'related' includes related groups (stops at sRAG_MAX_FILES(environment variable, not marked required): Maximum number of files to keep in search results. Results are filtered to include only chunks from the top N best-scoring files. For example, 1 returns only thCHUNK_MIN_LENGTH(environment variable, not marked required): Minimum chunk length in characters (1-10000, defaults to 50). Chunks shorter than this threshold are filtered out during ingestion.STORE_IMAGES(environment variable, not marked required): Store supported PDF and DOCX images during ingestion and return them with matched chunks (defaults to false).EMBED_TITLE_PREFIX(environment variable, not marked required): Embed each chunk together with its document title, which can help when passages don't restate the topic the title names (defaults to false). After changing it,EMBED_HEADING_PREFIX(environment variable, not marked required): Add section headings to chunk embeddings when they fit (defaults to false). Independent of EMBED_TITLE_PREFIX. Re-ingest documents after changing it.RAG_DEVICE(environment variable, not marked required): Execution device for the embedder (defaults to cpu). Passed straight to ONNX Runtime; see the Transformers.js device source for the supported backend names. IfRAG_DTYPE(environment variable, not marked required): Embedding quantization dtype for the embedder (defaults to fp32). Opt-in and pass-through; accepts any dtype the chosen model provides (fp32, fp16, q8, int8, ..RAG_HYBRID_WEIGHT(environment variable, not marked required): Keyword boost factor for hybrid search (0.0-1.0, defaults to 0.6). 0 means semantic similarity only; higher values increase the keyword-match contribution to thRAG_RERANK_CMD(environment variable, not marked required): External reranker command template. Use {query} and {top} for query text and result count; unset disables reranking.RAG_RERANK_TIMEOUT_MS(environment variable, not marked required): Time budget per rerank call in milliseconds (100-600000, defaults to 10000). On timeout the spawned command is killed and the pre-rerank ordering is returned.
Links
RepositoryRegistry record (JSON)npm package
Being listed is not a safety check. Read the repository and check what a server asks for before you connect it. How the directory works.
Ratings and reviews
No reviews yet.
Reviews are not open yet. The review inbox is still being set up, so this page cannot take a review today. Nothing is shown here until a real user has written one.