How do I install MCP Local Rag?

Copy the stdio config for Cursor, Claude Desktop, Claude Code, or VS Code on this page. It uses the publisher's install, not a PlugMCP proxy.

Does MCP Local Rag need an API key?

The publisher lists these environment variables: BASE_DIR, BASE_DIRS, DB_PATH, CACHE_DIR, MODEL_NAME, MAX_FILE_SIZE, RAG_MAX_DISTANCE, RAG_GROUPING, RAG_MAX_FILES, CHUNK_MIN_LENGTH, RAG_DEVICE, RAG_DTYPE, RAG_HYBRID_WEIGHT, RAG_WATCH, HF_ENDPOINT, HF_AUTO_MIRROR, HTTPS_PROXY, HTTP_PROXY.

What transport does MCP Local Rag use?

MCP Local Rag runs locally over stdio.

When was MCP Local Rag last checked?

Last checked 2026-11-09. Semantic code & doc search with keyword boost. AST code nav, auto HF mirror, local, privacy-first.

MCP Local Rag MCP server

Power your agent

Add this server

The publisher's own command. Not a PlugMCP endpoint.

Add to Cursor Add to VS Code Run on PlugMCP

Host a copy with our keys and logs.

Local command

npx -y @damoqiongqiu/[email protected]

Credentials from the publisher

BASE_DIR — Base directory for document storage (defaults to current working directory). Ignored when BASE_DIRS is set.
BASE_DIRS — JSON array of base directories (e.g. '["/a","/b"]'). Takes precedence over BASE_DIR.
DB_PATH — Path to LanceDB database directory (defaults to ./lancedb/)
CACHE_DIR — Directory where Transformers.js models are cached (defaults to ./models/)
MODEL_NAME — Embedding model name (defaults to Xenova/all-MiniLM-L6-v2)
MAX_FILE_SIZE — Maximum file size in bytes (defaults to 104857600 / 100MB)
RAG_MAX_DISTANCE — Maximum distance threshold for filtering search results. Results with distance greater than this value will be excluded. Lower values mean stricter filtering (e.g., 0.5 for high relevance only)
RAG_GROUPING — Grouping mode for quality filtering. 'similar' returns only the most similar group (stops at first distance jump). 'related' includes related groups (stops at second distance jump). Unset means no grouping filter
RAG_MAX_FILES — 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 the single best-matching file's chunks. Unset means no file filtering.
CHUNK_MIN_LENGTH — Minimum chunk length in characters (1-10000, defaults to 50). Chunks shorter than this threshold are filtered out during ingestion.
RAG_DEVICE — Execution device for the embedder (defaults to cpu). Passed straight to ONNX Runtime; see the Transformers.js device source for the supported backend names. If the requested device fails to initialize, the server throws an error.
RAG_DTYPE — 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, ...). If the model has no variant for the requested dtype, the server throws an error. Changing this changes the embedding space — re-ingest existing data.
RAG_HYBRID_WEIGHT — 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 the final score.
RAG_WATCH — Enable automatic file watcher to detect changes and re-index modified files (true/1 to enable). Useful during active development.
HF_ENDPOINT — Explicit HuggingFace endpoint URL to use for model downloads. When set, auto-mirror detection is skipped and this URL is used directly. Useful when you have a full-featured mirror.
HF_AUTO_MIRROR — Enable automatic mirror detection for HuggingFace downloads (defaults to true). Set to "false" or "0" to disable and use huggingface.co directly. The mirror chain is: huggingface.co → hf-mirror.com → modelscope.cn.
HTTPS_PROXY — HTTPS proxy URL for downloading embedding models (e.g. http://proxy:8080). Required only if behind a corporate proxy.
HTTP_PROXY — HTTP proxy URL for downloading embedding models. Fallback if HTTPS_PROXY is not set.

Claude Code has no install link for a local command. Copy the command below and run it in a terminal.

shClaude Code

1BASE_DIR=<value> BASE_DIRS=<value> DB_PATH=<value> CACHE_DIR=<value> MODEL_NAME=<value> MAX_FILE_SIZE=<value> RAG_MAX_DISTANCE=<value> RAG_GROUPING=<value> RAG_MAX_FILES=<value> CHUNK_MIN_LENGTH=<value> RAG_DEVICE=<value> RAG_DTYPE=<value> RAG_HYBRID_WEIGHT=<value> RAG_WATCH=<value> HF_ENDPOINT=<value> HF_AUTO_MIRROR=<value> HTTPS_PROXY=<value> HTTP_PROXY=<value> claude mcp add io.github.damoqiongqiu-mcp-local-rag -- npx -y @damoqiongqiu/mcp-local-rag@0.21.0

Run in a terminal.

jsonClaude Desktop

1{2  "mcpServers": {3    "io.github.damoqiongqiu-mcp-local-rag": {4      "command": "npx",5      "args": [6        "-y",7        "@damoqiongqiu/[email protected]"8      ],9      "env": {10        "BASE_DIR": "<value>",11        "BASE_DIRS": "<value>",12        "DB_PATH": "<value>",13        "CACHE_DIR": "<value>",14        "MODEL_NAME": "<value>",15        "MAX_FILE_SIZE": "<value>",16        "RAG_MAX_DISTANCE": "<value>",17        "RAG_GROUPING": "<value>",18        "RAG_MAX_FILES": "<value>",19        "CHUNK_MIN_LENGTH": "<value>",20        "RAG_DEVICE": "<value>",21        "RAG_DTYPE": "<value>",22        "RAG_HYBRID_WEIGHT": "<value>",23        "RAG_WATCH": "<value>",24        "HF_ENDPOINT": "<value>",25        "HF_AUTO_MIRROR": "<value>",26        "HTTPS_PROXY": "<value>",27        "HTTP_PROXY": "<value>"28      }29    }30  }31}

claude_desktop_config.json

Local · npmPublisher installStep-by-step: add this in Claude

Remote URL or a command on your machine? How the two installs differ

Tools

No tool list yet. Remote servers get one after the first health check.

FAQ

Questions

MCP Local RAG

🍴 Forked from shinpr/mcp-local-rag — original work by Shinsuke Kagawa

Local code intelligence engine for AI coding assistants. AST-level semantic chunking + keyword boost for pinpointing functions, classes, and APIs — fully private, zero setup.

📖 中文文档

Features
Quick Start
- 2.1 Configure Your AI Coding Tool
- 2.2 CLI Quick Start
- 2.3 First-Time Project Indexing
Core Concepts
- 3.1 Dual-Strategy Chunking
- 3.2 Hybrid Search
- 3.3 Security Boundary
MCP Tool Reference
- 4.1 Ingest Tools
- 4.2 Search Tools
- 4.3 Management Tools
- 4.4 Code Intelligence
- 4.5 System Tools
CLI
- 5.1 Basic Commands
- 5.2 CLI Configuration
Network & Models
- 6.1 Mirror Auto-Detection
- 6.2 Model Selection
- 6.3 File Watching
Search Tuning
Performance Tuning
- 8.1 Quantization Precision
- 8.2 Execution Device
- 8.3 Minimum Chunk Length
- 8.4 Recommended Configurations

1. Features

Smart dual-strategy chunking — AST-level code chunking via tree-sitter (splits at function/class/method boundaries, injects scope chain + imports). Semantic chunking for documents (splits by meaning, not character count).
Semantic search + keyword boost — Vector search first, then keyword matching boosts exact terms. useEffect, error codes, class names rank higher — not just semantically guessed.
15 MCP tools — Ingest, search, manage, code intelligence, and system ops in one server.
AST code intelligence — find_definition and find_references for IDE-level code navigation, powered by tree-sitter metadata captured at ingest time.
Three-tier mirror auto-fallback — huggingface.co → hf-mirror.com → modelscope.cn, zero config for users in mainland China.
Runs entirely locally — No API keys, no cloud, no data leaving your machine. Works offline after the first model download.
Zero-friction setup — One npx command. No Docker, Python, or servers to manage.

2. Quick Start

Set BASE_DIR to the folder you want to search (BASE_DIRS for multiple roots — see Configuration).

2.1 Configure Your AI Coding Tool

Cursor — ~/.cursor/mcp.json:

{
  "mcpServers": {
    "local-rag": {
      "command": "npx",
      "args": ["-y", "@damoqiongqiu/mcp-local-rag"],
      "env": { "BASE_DIR": "/path/to/your/project" }
    }
  }
}

Claude Code:

claude mcp add local-rag --scope user --env BASE_DIR=/path/to/your/project -- npx -y @damoqiongqiu/mcp-local-rag

Codex — ~/.codex/config.toml:

[mcp_servers.local-rag]
command = "npx"
args = ["-y", "@damoqiongqiu/mcp-local-rag"]

[mcp_servers.local-rag.env]
BASE_DIR = "/path/to/your/project"

WorkBuddy — Settings → Custom Connectors → Add:

{
  "mcpServers": {
    "local-rag": {
      "command": "npx",
      "args": ["-y", "@damoqiongqiu/mcp-local-rag"],
      "env": { "BASE_DIR": "/path/to/your/project" }
    }
  }
}

⚠️ WorkBuddy: you MUST click "Trust" in the Custom Connectors list after adding, otherwise the server is silently blocked.

2.2 CLI Quick Start

No MCP needed — run directly from the terminal:

npx @damoqiongqiu/mcp-local-rag ingest ./src/
npx @damoqiongqiu/mcp-local-rag query "auth middleware"
npx @damoqiongqiu/mcp-local-rag status

That's it. No Docker, Python, or server setup.

2.3 First-Time Project Indexing

You: "Index the src directory of this project"
Assistant: Successfully ingested 156 files (2,847 chunks created)

You: "Where's the middleware that handles API rate limiting?"
Assistant: src/middleware/rateLimiter.ts — useRateLimiter(), lines 42–89

You: "How is the database connection pool configured?"
Assistant: src/config/database.ts — createPool() default max: 20, idle: 5

3. Core Concepts

3.1 Dual-Strategy Chunking

Chunking strategy is chosen per file type:

Code files (50+ languages) — CodeChunker parses source via tree-sitter AST, splits at structural boundaries (functions, classes, methods). Each chunk's contextualizedText includes its scope chain and import context for precise semantic search.
Documents (PDF/DOCX/TXT/MD/HTML) — SemanticChunker splits into sentences, groups by embedding similarity to find natural topic boundaries. Markdown code blocks remain intact — never split mid-block.

3.2 Hybrid Search

Search = semantic similarity + keyword boost (RAG_HYBRID_WEIGHT, default 0.6):

Query vectorization → semantic search finds most relevant chunks
Quality filters apply (distance threshold, grouping)
Keyword matching boosts exact-term rankings

Exact identifiers like useEffect are never buried by semantic approximations.

3.3 Security Boundary

Only files under BASE_DIR / BASE_DIRS are accessible for ingest, list, delete, or read-neighbor operations. Symlinks resolved outside roots are rejected. Sibling-prefix paths (e.g., /foo/barista when root is /foo/bar) are also blocked — prevents path traversal attacks.

4. MCP Tool Reference

15 tools organized into 5 categories.

4.1 Ingest Tools

#	Tool	Purpose	Example
1	`ingest_file`	Single file (PDF/DOCX/TXT/MD/code)	`"Ingest ./docs/api-spec.pdf"`
2	`ingest_data`	In-memory text/HTML	`"Fetch this page and ingest the HTML"`
3	`ingest_directory`	Bulk directory ingest	`"Ingest everything under ./src"`

ingest_file supports 50+ code languages. PDFs support an optional visual mode — a local VLM generates captions for figure pages, making visual content searchable. Two profiles available:

Profile	Model	Cache	Suited for
`fast` (default)	SmolVLM-256M	~250 MB	Light visual indexing
`quality`	Qwen2.5-VL-3B-ONNX	~2.9 GB	Figures with in-image text

# CLI
npx @damoqiongqiu/mcp-local-rag ingest ./spec.pdf --visual --visual-quality quality
# MCP
"Ingest ./spec.pdf with visual: true, visualQuality: 'quality'"

ingest_data runs Readability → Markdown → index. Perfect for web content fetched by your AI assistant. Re-ingesting replaces old versions automatically.

ingest_directory scans recursively, respects .gitignore, shows real-time progress via MCP notifications.

4.2 Search Tools

#	Tool	Purpose	Key Parameters
4	`query_documents`	Hybrid search (semantic + keyword)	`query`, `limit`, `scope`, `highlightContext`, `fromTimestamp`
5	`read_chunk_neighbors`	Expand context around results	`filePath`, `chunkIndex`, `before`, `after`

query_documents — scope accepts a single path prefix or list, restricting results to that subtree. highlightContext returns snippets around matched terms. fromTimestamp / untilTimestamp enable time-range filtering.

read_chunk_neighbors — defaults to 2 chunks before and after (like grep -C 2), max 50 each. Response includes the target chunk marked isTarget: true.

4.3 Management Tools

#	Tool	Purpose
6	`list_files`	List files with ingestion status (`ingested: true/false`)
7	`delete_file`	Delete by file path or source URL
8	`status`	Index stats: docs, chunks, memory, search mode

list_files supports scope filtering with the same prefix-match semantics as search. In large directories, scope accelerates the scan by skipping out-of-scope subtrees.

4.4 Code Intelligence

#	Tool	Purpose	Input
9	`find_definition`	Locate symbol definition (file, line range, scope)	Exact symbol name
10	`find_references`	Find all references (import + text mention)	Symbol name

Both tools depend on AST metadata (imports, entities, scope chains) extracted by tree-sitter at ingest time. Only works for code files ingested with CodeChunker — files ingested before v0.18.7 lack this metadata and require reindex_all to rebuild.

find_references uses a two-phase strategy: (1) exact match in codeMeta.imports → (2) FTS full-text search for the symbol name. Results are deduplicated by (filePath, chunkIndex), with import references listed first.

4.5 System Tools

#	Tool	Purpose
11	`config`	Runtime hot read/write config — no restart needed
12	`dedup_check`	SHA256 + Jaccard similarity to detect duplicate files
13	`export_index`	Export entire index as JSON (backup or migration)
14	`reindex_all`	Full re-chunk + re-embed (after model change)
15	`reindex_stale`	Re-ingest only files modified on disk (incremental sync)

config hot-swaps hybridWeight, modelName, cacheDir, baseDir/baseDirs, etc. Switching models auto-disposes the old Embedder and initializes the new one — note: changing models alters the embedding space and requires reindex_all.

dedup_check is especially useful in monorepos — spot ↔ futures mirror code is typically flagged with similarity 1.0.

5. CLI

5.1 Basic Commands

# Ingest
npx @damoqiongqiu/mcp-local-rag ingest ./src/

# Search (with scope)
npx @damoqiongqiu/mcp-local-rag query "auth middleware"
npx @damoqiongqiu/mcp-local-rag query "auth" --scope /docs/api

# Context expansion
npx @damoqiongqiu/mcp-local-rag read-neighbors --file-path /abs/path.md --chunk-index 5

# Management
npx @damoqiongqiu/mcp-local-rag list --scope /docs/api
npx @damoqiongqiu/mcp-local-rag status
npx @damoqiongqiu/mcp-local-rag delete ./docs/old.pdf
npx @damoqiongqiu/mcp-local-rag delete --source "https://..."

query, read-neighbors, list, status, delete emit JSON to stdout (pipe to jq). ingest emits progress to stderr.

Global options (--db-path, --cache-dir, --model-name) go before the subcommand:

npx @damoqiongqiu/mcp-local-rag --help

⚠️ The CLI does NOT read your MCP client config (mcp.json, etc.). Configure via flags or environment variables.

5.2 CLI Configuration

Flags — global options before, subcommand options after:

npx @damoqiongqiu/mcp-local-rag --db-path ./my-db query "auth" --base-dir ./docs

--base-dir is repeatable on ingest and list:

npx @damoqiongqiu/mcp-local-rag ingest --base-dir ./docs --base-dir ./specs ./docs/readme.md

Environment variables:

export DB_PATH=./my-db
export BASE_DIR=./docs
npx @damoqiongqiu/mcp-local-rag query "auth"

For multiple roots, use BASE_DIRS (JSON array):

export BASE_DIRS='["/Users/me/work","/Users/me/specs"]'

Precedence: CLI flags > environment variables > defaults.

6. Network & Models

6.1 Mirror Auto-Detection

huggingface.co is inaccessible from mainland China. Built-in three-tier mirror chain with automatic fallback:

huggingface.co → hf-mirror.com → modelscope.cn

At startup, each mirror is HEAD-probed (3s timeout). The first reachable mirror with a complete API is selected:

With proxy (HTTPS_PROXY) → direct to huggingface.co
No proxy → auto-switch to hf-mirror.com
hf-mirror API unavailable → fallback to modelscope.cn

No manual HF_ENDPOINT required. For manual control:

Env Var	Effect
`HF_AUTO_MIRROR=false`	Disable auto-detection, use huggingface.co only
`HF_ENDPOINT=<url>`	Force a specific mirror, skip auto-detection

v0.18.5+ uses setGlobalDispatcher(ProxyAgent) — all Node.js 22 network requests go through the proxy.

6.2 Model Selection

6 embedding models with alias resolution via model-registry:

Model	Alias	Size	Dims
`Xenova/all-MiniLM-L6-v2` (default)	`mini`	~90 MB	384
`Xenova/all-MiniLM-L12-v2`	—	~120 MB	384
`Xenova/bge-small-en-v1.5`	`bge-small`	~130 MB	384
`Xenova/all-mpnet-base-v2`	`mpnet`	~420 MB	768
`Xenova/bge-base-en-v1.5`	—	~420 MB	768
`Xenova/multi-qa-mpnet-base-dot-v1`	`multi-qa`	~420 MB	768

Guidance: code repos → default model + high keyword boost; multilingual → consider embeddinggemma-300m; scientific papers → consider allenai-specter.

RAG_DTYPE controls ONNX precision (fp32 / fp16 / q8). Default fp32; use q8 when memory-constrained. ⚠️ Changing models or dtype requires deleting DB_PATH and re-indexing.

6.3 File Watching

Set RAG_WATCH=true — the server starts recursive fs.watch on baseDirs (500ms debounce):

File creation/modification → auto ingest_file
File deletion → auto delete_file

Ideal for actively changing projects.

7. Search Tuning

Variable	Default	Description
`RAG_HYBRID_WEIGHT`	`0.6`	Keyword boost: 0 = semantic only, 1 = keyword only
`RAG_GROUPING`	unset	`similar` = top group only, `related` = top 2 groups
`RAG_MAX_DISTANCE`	unset	Filter low-relevance results (e.g., `0.5`)
`RAG_MAX_FILES`	unset	Limit results to top N files

Code-focused tuning (recommended default):

{ "RAG_HYBRID_WEIGHT": "0.7", "RAG_GROUPING": "similar" }

Document-focused tuning:

{ "RAG_HYBRID_WEIGHT": "0.4", "RAG_GROUPING": "related" }

Keyword boost is applied after semantic filtering — improves precision without introducing noise.

8. Performance Tuning

Beyond search accuracy, inference performance is also configurable. All optimizations are environment variables — no code changes required.

8.1 Quantization Precision (`RAG_DTYPE`)

Controls ONNX model inference precision. For all-MiniLM-L6-v2, three levels are available:

Value	Model Size	Speed	Memory	Precision Loss	Best For
`fp32` (default)	~90 MB	baseline	~80 MB	none	First use, maximum accuracy
`fp16`	~45 MB	20-30% faster	~45 MB	negligible	Recommended for daily use
`q8`	~45 MB	30-50% faster	~45 MB	minor	Low memory, large projects

"env": { "RAG_DTYPE": "fp16", "BASE_DIR": "..." }

⚠️ Changing dtype requires index rebuild — embedding spaces are incompatible.

Verify it works: After restart, call status via MCP and check the dtype field. Should match your setting (e.g., "fp16").

If it fails: Startup throws EmbeddingError with a list of supported dtypes. Common cause: the model doesn't provide the q8 variant — switch to fp16.

8.2 Execution Device (`RAG_DEVICE`)

Controls which ONNX Runtime backend to use:

Value	Backend	Notes
`cpu` (default)	CPU	Most stable, no extra dependencies
`webgpu`	GPU (WebGPU)	⚠️ Experimental: M1/M2 Mac uses Metal, NVIDIA uses Vulkan

"env": { "RAG_DEVICE": "webgpu", "RAG_DTYPE": "fp16", "BASE_DIR": "..." }

⚠️ Changing device changes the embedding space — requires index rebuild. Stacks with RAG_DTYPE — fp16 + webgpu gives both model-size reduction and GPU speedup.

Verify it works: MCP startup log should show Loading model on device "webgpu". status should show device: "webgpu".

If it fails:

Unsupported device at startup → WebGPU unavailable in your environment, revert to "cpu"
Starts successfully but inference crashes → likely an ONNX WebGPU backend bug, revert to "cpu"
Just delete the RAG_DEVICE line to fall back — other config is untouched

8.3 Minimum Chunk Length (`CHUNK_MIN_LENGTH`)

Filters out chunks shorter than this value during ingest. Default 50 keeps nearly everything; 200 drops 30-40% of noise fragments.

"env": { "CHUNK_MIN_LENGTH": "200", "BASE_DIR": "..." }

⚠️ Blunt instrument — short but important code (e.g., config constants) may also be discarded. Requires index rebuild. Sweet spot: 100-200.

8.4 Recommended Configurations

Scenario	Config
Daily development	`RAG_DTYPE=fp16`
Large project + M1/M2 Mac	`RAG_DTYPE=fp16, RAG_DEVICE=webgpu`
Memory-constrained	`RAG_DTYPE=q8`

All changes require reindex_all (MCP) or re-running ingest (CLI). If something breaks, delete the failing env line to revert to defaults.

9. Configuration Reference

MCP server: environment variables only (via your MCP client's env block). CLI: environment variables + equivalent flags (flags take precedence).

Env Var	CLI Flag	Default	Description
`BASE_DIR`	`--base-dir` (repeatable)	`cwd`	Document root (security boundary)
`BASE_DIRS`	—	unset	JSON array of roots, overrides `BASE_DIR`
`DB_PATH`	`--db-path`	`./lancedb/`	Vector database path
`CACHE_DIR`	`--cache-dir`	`./models/`	Model cache — recommend absolute path
`MODEL_NAME`	`--model-name`	`all-MiniLM-L6-v2`	HuggingFace model ID
`MAX_FILE_SIZE`	`--max-file-size`	100 MB	Max file size in bytes
`CHUNK_MIN_LENGTH`	`--chunk-min-length`	`50`	Min chunk length (1–10000 chars)
`RAG_DEVICE`	—	`cpu`	ONNX execution device
`RAG_DTYPE`	—	`fp32`	Quantization (`fp32`/`fp16`/`q8`)
`HTTPS_PROXY`	—	unset	Model download proxy. v0.18.5+ globally effective
`HF_ENDPOINT`	—	`huggingface.co`	Manual mirror override
`HF_AUTO_MIRROR`	—	`true`	Auto-detection toggle
`RAG_WATCH`	—	unset	File watching (`true`/`1`)

Root resolution order: CLI --base-dir > BASE_DIRS > BASE_DIR > cwd. BASE_DIRS and BASE_DIR are never merged. Only JSON array syntax supported for BASE_DIRS — delimiter syntax is intentionally rejected.

10. Troubleshooting

Symptoms: fetch failed, status shows searchMode: fts instead of hybrid.

Solutions:

Network restriction (mainland China, etc.) — use proxy:
```
"env": { "HTTPS_PROXY": "http://127.0.0.1:7890" }
```
Set in your MCP client config, not the terminal. v0.18.5+ globally effective via setGlobalDispatcher.
Auto-mirror fallback (v0.18.2+, default) — three-tier probe. Usually works without any config.
Manual override — HF_ENDPOINT=https://modelscope.cn or download models manually into CACHE_DIR.
npx cached old version — clear and restart:
```
rm -rf ~/.npm/_npx/
```

Verify config file syntax
WorkBuddy users: confirm "Trust" button clicked
Restart client completely (Cmd+Q on macOS)
Test directly: npx @damoqiongqiu/mcp-local-rag should run without errors

After switching models or when the database is corrupted:

Stop the MCP service
Delete DB_PATH directory (default ./lancedb/) — safe, doesn't affect source files
Restart MCP → fresh database auto-created

Bulk re-ingest:

npx @damoqiongqiu/mcp-local-rag ingest ./src/

Private? Yes. After model download, nothing leaves your machine.
Offline? Yes, once models are cached.
Supported formats? 50+ code languages + PDF/DOCX/TXT/MD/HTML. No Excel, PPT, or images.
GPU acceleration? Opt-in via RAG_DEVICE. Support depends on your system, Node.js version, and the ONNX backend.
Backup? Copy the DB_PATH directory.

11. Development

git clone https://github.com/damoqiongqiu/mcp-local-rag.git
cd mcp-local-rag
pnpm install

pnpm test              # All tests
pnpm run type-check    # TypeScript check
pnpm run check:fix     # Lint + format
pnpm run check:all     # Full CI pipeline

src/
  index.ts      # Entry point
  server/       # MCP tool handlers
  cli/          # CLI subcommands
  parser/       # PDF/DOCX/TXT/MD/code parsing
  chunker/      # SemanticChunker + CodeChunker
  embedder/     # Transformers.js embeddings
  vectordb/     # LanceDB operations
  utils/        # Shared utilities (security, scan, scope)
  __tests__/    # Test suites

License

MIT License. Free for personal and commercial use.

Acknowledgments

Built with Model Context Protocol (Anthropic), LanceDB, and Transformers.js.