ModelsAtlas
Explore
Atlas / MCP servers / shemhamforash23 / LightRAG

LightRAGCAUTION

mcp/shemhamforash23/lightrag
MCP server by shemhamforash23 · Audited 2026-10-07
⚑ Report / re-scanSource ↗
Verdict
CAUTION
Grade
B
Trust score
88 /100
Exposed tools
9 4r · 3w · 2d
Transport
streamable-http
License
—
Stars
123
01

Overview

From the repository's own README, as read at the audited commit. Badges and raw HTML are left out.

[](https://mseep.ai/app/shemhamforash23-lightrag-mcp)

MCP server for integrating LightRAG with AI tools. Provides a unified interface for interacting with LightRAG API through the MCP protocol.

Description

LightRAG MCP Server is a bridge between LightRAG API and MCP-compatible clients. It allows using LightRAG (Retrieval-Augmented Generation) capabilities in various AI tools that support the MCP protocol.

Key Features

  • Information Retrieval: Execute semantic and keyword queries to documents
  • Document Management: Upload, index, and track document status
  • Knowledge Graph Operations: Manage entities and relationships in the knowledge graph
  • Monitoring: Check LightRAG API status and document processing

Installation

This server is designed to be used as an MCP server and should be installed in a virtual environment using uv, not as a system-wide package.

Development Installation

# Create a virtual environment
uv venv --python 3.11

# Install the package in development mode
uv pip install -e .

Requirements

  • Python 3.11+
  • Running LightRAG API server

Usage

LightRAG MCP server supports two MCP transport modes:

  • stdio (default): run through an MCP client configuration file (mcp-config.json)
  • streamable-http: run as a standalone HTTP server for remote MCP clients

Command Line Options

LightRAG API connection:

  • --host: LightRAG API host (default: localhost)
  • --port: LightRAG API port (default: 9621, or 443 when --ssl is set)
  • --api-key: LightRAG API key (optional)
  • --ssl: Use HTTPS instead of HTTP (default: False)

MCP transport:

  • --mcp-transport: MCP transport (stdio or streamable-http, default: stdio)
  • --mcp-host: MCP HTTP host (default: 127.0.0.1)
  • --mcp-port: MCP HTTP port (default: 8000)
  • --mcp-streamable-http-path: Streamable HTTP endpoint
Read from source at commit 44ddf182323fOBSERVED · 2026-10-07
02

Connect

Built from this server's own package name, version and transport as found in its source — not copied from anyone's documentation, so it cannot drift against a page we do not control.

claude-code
claude mcp add lightrag_mcp -- uvx lightrag_mcp
claude-desktop
{
  "mcpServers": {
    "lightrag_mcp": {
      "command": "uvx",
      "args": [
        "lightrag_mcp"
      ]
    }
  }
}
03

Exposed tools (9)

4 read · 3 write · 2 destructive. Blast radius: 2 tools can delete or overwrite — an agent that can be talked into calling a tool can be talked into calling this one.

ToolRiskDescription
check_lightrag_healthreadasync def _operation(client: LightRAGClient) -> Any:
delete_by_doc_idsdestructive
delete_by_entitiesdestructive
get_documentsreadasync def _operation(client: LightRAGClient) -> Any:
get_pipeline_statusreadasync def _operation(client: LightRAGClient) -> Any:
insert_documentwriteasync def _operation(client: LightRAGClient) -> Any:
insert_filewriteasync def _operation(client: LightRAGClient) -> Any:
scan_for_new_documentsreadasync def _operation(client: LightRAGClient) -> Any:
upload_documentwriteasync def _operation(client: LightRAGClient) -> Any:
04

Trust audit

CAUTIONgrade B · trust 88/100 Install with care. The audit found things worth knowing before you trust its output.

LayerWhat it checksResult
L0Provenance & inventoryPASS
L1Static analysis of the codePASS
L2Instruction surface (what it tells the agent)PASS
L3Class-specific surfaceWARN
L4Behavioural (sandbox)SKIPPED

What the source does

Filesystem
declared (1 observation(s))
Network
declared (5 observation(s))
Shell
none-observed
Dependencies
pinned
Secrets in source
none-found

Findings (3)

MEDIUMFilesystem / path · mcp.destructive_tools · CWE-22, CWE-59
delete_by_doc_ids, delete_by_entities
Why it matters. 2 tool(s) can delete or overwrite
Fix. prefer a read-only mode or scoped tokens; the page states the blast radius
MEDIUMAuth / authz · mcp.remote_no_auth · CWE-287, CWE-862
streamable-http
Why it matters. a network transport with no auth environment variable found
Fix. require a token
LOWInventory / provenance · inv.no_license · CWE-1104
Why it matters. no LICENSE file and no repo licence
Fix. add a licence

Gates applied: no_behavioural_pass, no_license.

Audited 2026-10-07 · audit v0.4.1 · source sha 44ddf182323ffull audit observations/trust-audit/mcp-server/shemhamforash23__lightrag.json · Report an issue / request a re-scan
05

Audit history

Every audit this server has had. A grade with a past is a grade somebody is still checking.

DateSourceVerdictGradeScoreChange
2026-10-0744ddf182323fCAUTIONB88first audit
06

Questions

What tools does LightRAG expose?

9 in total: 4 read-only, 3 that write, and 2 that can delete or overwrite (delete_by_doc_ids, delete_by_entities). Every one is listed on this page with its risk.

Is LightRAG safe to connect to an agent?

With care. The audit graded it B (88/100) and found 3 things worth knowing before you trust this server, listed below with the exact line each was found on. Separately from the audit: 2 of its tools can destroy data, so scope the token you give it to what you actually need.

What credentials does LightRAG need?

No credential environment variables were found in its source, so it appears to need none.

How does LightRAG run?

It speaks streamable-http, so it runs as a service you connect to over the network. It is published on PyPI as lightrag_mcp.

How current is this page?

The grade is for one exact copy of the source (44ddf182323f), read on 2026-10-07. The repository is watched and re-audited when it changes.

Advertisement