Atlas / MCP servers / grll / Mcpadapt

McpadaptBLOCK

Name: Mcpadapt
Author: grll

mcp/grll/mcpadapt

MCP server by grll · langchain · llamaindex · modelcontextprotocol · Audited 2026-10-01

⚑ Report / re-scan Source ↗

Unlock 650+ MCP servers tools in your favorite agentic framework.

Verdict

BLOCK

Grade

Trust score

69 /100

Exposed tools

18 18r · 0w · 0d

Transport

stdio · streamable-http

License

MIT

Stars

426

Overview

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

For more context, guides and API references visit our new documentation.

Unlock 650+ MCP servers tools in your favorite agentic framework.

Model Context Protocol is an open-source protocol introduced by Anthropic which allows anyone to simply and quickly make tools and resources available as "MCP Servers".

Since its release more than 650 MCP servers have been created giving access to many data & tools to supported "MCP Client".

This project makes calling any MCP servers tools seemless from any agentic framework. Virtually providing your agentic worfklow access to 650+ MCP servers tools.

Look at glama.ai or smithery.ai to give you an idea of what your agent could access.

Installation Instructions

Smolagents

Smolagents 1.4.1 and above directly ships with mcpadapt integrated in their tool collections object. It means you can directly use it from smolagents:

uv add smolagents[mcp]

Other Frameworks

Each agent framework has its own set of optional dependencies to not clutter with useless dependencies. You choose the flavor you want by adding your framework in brackets in the installation command.

# with uv
uv add mcpadapt[langchain]

# or with pip
pip install mcpadapt[langchain]

Framework supported at the moment: smolagents, langchain, crewAI, google-genai.

You can also add multi

Read from source at commit 23d82025f966OBSERVED · 2026-10-01

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 mcpadapt --env ANTHROPIC_API_KEY=${ANTHROPIC_API_KEY} --env GEMINI_API_KEY=${GEMINI_API_KEY} --env HF_TOKEN=${HF_TOKEN} --env OPENAI_API_KEY=${OPENAI_API_KEY} -- uvx mcpadapt

claude-desktop

{
  "mcpServers": {
    "mcpadapt": {
      "command": "uvx",
      "args": [
        "mcpadapt"
      ],
      "env": {
        "ANTHROPIC_API_KEY": "${ANTHROPIC_API_KEY}",
        "GEMINI_API_KEY": "${GEMINI_API_KEY}",
        "HF_TOKEN": "${HF_TOKEN}",
        "OPENAI_API_KEY": "${OPENAI_API_KEY}"
      }
    }
  }
}

Exposed tools (18)

18 read · 0 write · 0 destructive.

Tool	Risk	Description
`custom_tool`	read	pass
`def`	read	Echo the input text
`dict_tool`	read	Returns a dictionary
`echo-tool`	read	Echo the input text
`echo_tool`	read	Echo the input text
`echo_tool_default_value`	read	Echo the input text, default to
`echo_tool_optional`	read	Echo the input text, or return a default message if no text is provided
`echo_tool_union_none`	read	Echo the input text, but None is not specified by default.
`get_product_info`	read	Get detailed information about a product
`invalid_tool`	read	Tool that returns invalid JSON when dict is expected.
`list_tool`	read	Returns a list
`multi_type_tool`	read	Test tool with a parameter that accepts multiple types using array notation
`multiple_content_tool`	read	A tool that returns multiple content items
`problematic_tool`	read	Tool with empty union after null filtering
`strict_tool`	read	A tool that expects only string values
`string_tool`	read	Returns a string
`test_audio`	read	path = os.path.join(
`test_image`	read	path = os.path.join(

Trust audit

BLOCKgrade D · trust 69/100 Do not install this without reading the findings. The audit found something that could harm you or your machine.

Layer	What it checks	Result
L0	Provenance & inventory	PASS
L1	Static analysis of the code	FAIL
L2	Instruction surface (what it tells the agent)	PASS
L3	Class-specific surface	PASS
L4	Behavioural (sandbox)	SKIPPED

What the source does

Filesystem: none-observed
Network: declared (1 observation(s))
Shell: declared (3 observation(s))
Dependencies: pinned
Secrets in source: none-found

Findings (15)

HIGHCode injection · code.eval_exec · CWE-78, CWE-94, CWE-95

src/mcpadapt/langchain_adapter.py:164

exec(generate_class_template(True), namespace)

Why it matters. evaluates text as code

Fix. remove; use a parser or a dispatch table

HIGHCode injection · code.eval_exec · CWE-78, CWE-94, CWE-95

src/mcpadapt/langchain_adapter.py:167

exec(generate_class_template(False), namespace)

Why it matters. evaluates text as code

Fix. remove; use a parser or a dispatch table

LOWInventory / provenance · inv.binary · CWE-1104

tests/data/white_noise.wav

white_noise.wav

Why it matters. a compiled or binary member cannot be reviewed from source

Fix. ship source, or explain the binary in the README

LOWInventory / provenance · inv.hidden_file · CWE-1104

.pre-commit-config.yaml

.pre-commit-config.yaml

Why it matters. hidden member outside the usual dotfiles

Fix. review its purpose