Loading ToolBench

Reading the current corpus…

padwan-ai MCP server · F · ToolBench

← Browse

padwan-ai

github.com/polarsen-io/padwan-ai

Lightweight async client for OpenAI, Gemini, Mistral, Grok, Anthropic and any OpenAI-compatible API.

2026-07-28+STDIO1 tools

20/100

Definition Quality5

Protocol Readiness0

Supportability57

Spec posture

Inferred effective spec 2026-07-28+·current baseline 2026-07-28

Static source inference · medium confidence

Current-spec patterns detected

Summary

This is not an MCP server. The repository is 'padwan-ai', a lightweight async LLM client library for OpenAI, Gemini, Mistral, Anthropic, etc. It contains NO MCP tool definitions, NO MCP server implementation, and NO MCP protocol support. The single 'tool' entry in the evaluation is a decorator/wrapper function from padwan_ai/tools.py that converts async Python functions into validated tools for use within the padwan-ai library itself, not MCP tools. The source shows tool-calling support for Anthropic and Gemini APIs (converting between provider formats), but no MCP server, no STDIO transport, no MCP protocol implementation. This is a category error: padwan-ai is an LLM client library, not an MCP server.

Tools (1)

toolread onlysource verified

An `McpTool` whose schema is `fn`'s typed signature; arguments are validated before `fn` runs. ``name`` defaults to the function name and ``description`` to its docstring. Every parameter needs a type annotation and must be passable by keyword; a default makes it optional. ``validator`` picks the library doing schema, validation and result dumping (``"msgspec"``, ``"pydantic"`` or any `ToolValidator`); when omitted, the library the annotations name is used, else the only installed one. Constraints and error text are the chosen library's own.

Issues (3)

CRITICALN/A ↗

Not an MCP server. padwan-ai is an LLM client library, not an MCP server implementation. No MCP protocol, no tool registration, no MCP transport.

tool

CRITICALtool ↗

The 'tool' entry is a Python decorator/wrapper for internal use, not an MCP tool definition. It has no input schema visible in the code, no parameter descriptions, and no output schema.

tool

CRITICALN/A ↗

No MCP server entry point, no MCP protocol handlers, no tool registration with MCP. The pyproject.toml lists 'mcp>=2.0.0' as a dev dependency but the library does not implement MCP.

N/A

Recommendations

If the goal is to create an MCP server: implement MCP protocol handlers (tools, resources, prompts), register tools with proper schemas and descriptions, and expose a STDIO or HTTP transport.
If padwan-ai is intended to be used as a library within an MCP server: document that clearly and provide example MCP server code that wraps padwan-ai's LLM clients.
Add explicit tool input schemas with parameter types, descriptions, and constraints if this is meant to be an MCP tool definition.
Provide output schema documentation for any tool that returns structured data.
If tool-calling is the feature, document the supported LLM providers and tool format conversions (Anthropic, Gemini, OpenAI, etc.) in a dedicated guide.

Score history

Overall score trend

First recorded score · v2 rubric

20/100

Scored	Grade	Overall	Spec posture	Rubric
2026-09-23	F	20	2026-07-28+	v2