ModelsAtlas
Explore
Atlas / MCP servers / jesse-merhi / RAG Anything

RAG AnythingSAFE

mcp/jesse-merhi/rag-anything
MCP server by jesse-merhi · Audited 2026-10-08
⚑ Report / re-scanSource ↗

An MCP server for rag-anything (A multi-modal RAG system)

Verdict
SAFE
Grade
B
Trust score
89 /100
Exposed tools
7 6r · 0w · 1d
Transport
stdio
License
MIT
Stars
37
01

Overview

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

An MCP (Model Context Protocol) server that provides comprehensive RAG (Retrieval-Augmented Generation) capabilities for processing and querying directories of documents using the raganything library with full multimodal support.

Features

  • End-to-End Document Processing: Complete document parsing with multimodal content extraction
  • Multimodal RAG: Support for images, tables, equations, and text processing
  • Batch Processing: Process entire directories with multiple file types
  • Advanced Querying: Both pure text and multimodal-enhanced queries
  • Multiple Query Modes: hybrid, local, global, naive, mix, and bypass modes
  • Vision Processing: Advanced image analysis using GPT-4V
  • Persistent Storage: RAG instances maintained per directory for efficient querying

Available Tools

process_directory

Process all files in a directory for comprehensive RAG indexing with multimodal support.

Required Parameters:

  • directory_path: Path to the directory containing files to process
  • api_key: OpenAI API key for LLM and embedding functions

Optional Parameters:

  • working_dir: Custom working directory for RAG storage
  • base_url: OpenAI API base URL (for custom endpoints)
  • file_extensions: List of file extensions to process (default: ['.pdf', '.docx', '.pptx', '.txt', '.md'])
  • recursive: Process subdirectories (default: True)
  • enable_image_processing: Enable image analysis (default: True)
  • enable_table_processing: Enable table extraction (default: True)
  • enable_equation_processing: Enable equation processing (default: True)
  • max_workers: Concurrent processing workers (default: 4)

process_single_document

Process a single document with full multimodal analysis.

Required Parameters:

  • file_path: Path to the document to process
  • api_key: OpenAI API key

Optional Parameters:

  • working_dir: Custom working directory for RAG storage
  • base_url: OpenAI API base U
Read from source at commit c091010be42aOBSERVED · 2026-10-08
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. Replace the environment placeholders with a token scoped to the least it needs.

claude-code
claude mcp add rag-anything-mcp --env OPENAI_API_KEY=${OPENAI_API_KEY} -- uvx rag-anything-mcp
claude-desktop
{
  "mcpServers": {
    "rag-anything-mcp": {
      "command": "uvx",
      "args": [
        "rag-anything-mcp"
      ],
      "env": {
        "OPENAI_API_KEY": "${OPENAI_API_KEY}"
      }
    }
  }
}
03

Exposed tools (7)

6 read · 0 write · 1 destructive. Blast radius: 1 tool can delete or overwrite — an agent that can be talked into calling a tool can be talked into calling this one.

ToolRiskDescription
check_doc_ingestedread
clear_all_datadestructive
get_workspace_inforeadget_workspace_info
process_directoryread
process_single_documentread
query_with_multimodalread
query_workspaceread
04

Trust audit

SAFEgrade B · trust 89/100 Nothing in the source contradicts what it says it does. Grade A is reserved for packages that have also passed the behavioural sandbox.

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 (2 observation(s))
Network
none-observed
Shell
none-observed
Dependencies
pinned
Secrets in source
none-found

Findings (1)

MEDIUMFilesystem / path · mcp.destructive_tools · CWE-22, CWE-59
clear_all_data
Why it matters. 1 tool(s) can delete or overwrite
Fix. prefer a read-only mode or scoped tokens; the page states the blast radius

Gates applied: no_behavioural_pass.

Audited 2026-10-08 · audit v0.4.1 · source sha c091010be42afull audit observations/trust-audit/mcp-server/jesse-merhi__rag-anything.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
2026-10-08c091010be42aSAFEB89first audit
06

Questions

What is the RAG Anything MCP server?

An MCP server for rag-anything (A multi-modal RAG system)

What tools does RAG Anything expose?

7 in total: 6 read-only, 0 that write, and 1 that can delete or overwrite (clear_all_data). Every one is listed on this page with its risk.

Is RAG Anything safe to connect to an agent?

The audit found nothing in the source that contradicts what it says it does, and graded it B (89/100). Grade A is held back for packages that have also passed a sandboxed behavioural run, which is why a clean server reads B. Separately from the audit: 1 of its tools can destroy data, so scope the token you give it to what you actually need.

What credentials does RAG Anything need?

It reads OPENAI_API_KEY from the environment. Give it a token scoped to the least it needs — an agent that can be talked into calling a tool can be talked into calling it with your credentials.

How does RAG Anything run?

It speaks stdio, so it runs as a local process your client starts. It is published on PyPI as rag-anything-mcp.

How current is this page?

The grade is for one exact copy of the source (c091010be42a), read on 2026-10-08. The repository is watched and re-audited when it changes.

Advertisement