Pare Python

Community

Structured Python tool output (ruff, mypy, pip, uv, black, pytest) as typed JSON.

How to install ↓ Compare GitHub source repository ↗

Category: Dev Tools #362 of 438
Stars: ★ 138 Popular
Transport: stdio (local process)
Runtime: Node.js 20+
Credentials: No credential needed
License: MIT
Last commit: 53 days ago

39FMRS · D

Pare Python provides a valuable MCP server for developers seeking structured Python tool output for AI workflows. It excels in token efficiency but may have dependency-related risks due to its nature as a third-party wrapper.

Strongest · Setup experience 11/20 Weakest · Reliability 6/20

Reliability

6/20

Security and permissions

6/20

Maintenance

8/20

Documentation

8/20

Setup experience

11/20

Why each score

Reliability 6/20: Evidence: the manifest is valid JSON, declares a single stdio package @paretools/python at version 0.22.0, and the README describes the Pare structuredContent/outputSchema architecture and lists ruff, mypy, pip, uv, black, pytest for this server. Deductions: no server-python source, tool list, unit/integration tests, or CI logs were supplied; repo-wide CI/codecov badges cannot be attributed to this specific server, so tool-behavior alignment, MCP init handshake, and error paths are unverifiable. Conservatively capped below 12; score 6.
Security and permissions 6/20: Evidence: the server description exposes a real risk surface by wrapping pip, uv, pytest, black, and ruff — commands that can execute third-party code or modify files; no credential or external-network data-flow documentation is present. Deductions: with no source, least-privilege handling, command confirmation, argument allow-listing, dangerous-operation confirmation, and data-flow disclosure cannot be verified; the README's PARE_TOOLS filtering only limits which tools are registered and is not a security control. No red-line issue (malware, real secrets, covert exfiltration) is visible in the supplied material, but security evidence is far from sufficient.
Maintenance 8/20: Evidence: the repository is not archived, an MIT license is declared, the manifest pins a concrete 0.22.0 version, and the README references CI/codecov/OpenSSF badges and a CONTRIBUTING link. Deductions: no commit history, release timeline, issue-response record, dependency-update policy, or security-response channel was provided; badges are repo-level and untrusted, so they do not prove sustained maintenance of the python server. Maintenance appears plausible but unproven; score 8.
Documentation 8/20: Evidence: the README provides general architecture docs, preset-based quick setup, per-client setup guides, tool-filtering env vars, and troubleshooting. Deductions: for the Python server specifically there is only a one-line description and a wrapper list; no per-tool parameters, output schemas, examples, limits, cost details, or Python-specific client config example are included. The tool-schemas links are not backed by supplied content, so coverage of this server cannot be confirmed. Hidden assumptions remain; score 8.
Setup experience 11/20: Evidence: the README shows an npx @paretools/init --preset python install path, lists client-specific setup guides, and the manifest routes the server through the npm package @paretools/python over stdio with Node >=20. Deductions: no Python-server-specific client config JSON example and no actual startup/CI verification for this package were supplied; the static calibration caps setup at 15 without verifiable execution evidence. The path is few-step but not fully evidenced, so score 11.

Static review · not runListed 2026-08-07

Read the FMRS scoring method →

Fit and risk

What it can accessReads local filesRuns commands or code

Best for

AI coding agents primarily working in the Python ecosystem.
Developers looking to save tokens when integrating development tools.
Scenarios requiring reliable integration with ruff, mypy, pip, uv, black, pytest.

Not for

Projects that do not use Python tools.
Tasks requiring interactive terminal sessions (e.g., certain pytest debug modes).
Environments without Node.js 20+ or npm.

Required permissions

Execute Python tools (ruff, mypy, pip, uv, black, pytest).
Access filesystem to run these tools.
Network access if pip/uv require it.

Risks and side effects

Executing arbitrary Python tools may have unintended side effects (e.g., installing packages).
Tool results may be inaccurate if environment paths are not correctly configured.
Dependency on external tool versions that may alter output formats.

Setup

Before you start

Runtime：Node.js 20+

Other optional settings (1)

PARE_TOOLS optional Universal tool filter in server:tool,tool format limiting tools registered across all servers, e.g. git:status,npm:install.

Add the following to your MCP client configuration:

{"mcpServers":{"pare-python":{"command":"npx","args":["-y","@paretools/python"]}}}

Alternatively, use npx @paretools/init --preset python for quick setup.

claude_desktop_config.json

{"mcpServers":{"pare-python":{"command":"npx","args":["-y","@paretools/python"]}}}

Shown for Claude Desktop. Other clients may use a different file or key (VS Code uses "servers") — the configurator below converts it.

.vscode/mcp.json

{
  "servers": {
    "pare-python": {
      "command": "npx",
      "args": [
        "-y",
        "@paretools/python"
      ]
    }
  }
}

Goes in your project's .vscode/mcp.json (VS Code uses a "servers" key).

Terminal

claude mcp add pare-python -- npx -y @paretools/python

Run it in a terminal; replace any <…> placeholders with your own values first.

Generate and validate client config

Check that it works

Add the pare-python server to your MCP client config and restart; its registered tools should appear in the tool list. Invoke a ruff or pytest-backed tool and confirm the response is structured JSON instead of raw text.

Troubleshooting

Ensure Node.js 20+ is installed: `node --version`.
Verify npx is in PATH; for Windows, use `cmd /c npx`.
Check that Python tools are installed and accessible in PATH.
If startup is slow, run `npx -y @paretools/python` once to populate the cache.

Things to try

Once connected, you can ask your AI assistant things like:

Run ruff on this project and give me the structured results
Run pytest and show me which tests failed
Type-check this repo with mypy
Format this repository with black

Use cases

Obtain stable, machine-readable output from Python developer tools for AI-driven coding tasks.

Reduce token usage by AI agents when processing command-line outputs.

Automate Python workflows (e.g., linting, type checking, testing) without parsing progress bars or ANSI escape codes.

Supported clients

Claude Desktop

Claude Code

Cursor

Listed from the project's documentation, not tested by this site.

Overview

Pare Python is an MCP server that wraps common Python development tools (ruff, mypy, pip, uv, black, pytest) and converts their command-line output into structured, schema-validated JSON. This eliminates the need to parse fragile terminal text, providing AI agents with reliable, typed data and reducing token consumption while improving workflow stability.

Similar servers

Context7 80 · B

Upstash's official server providing up-to-date third-party library docs for AI coding assistants

★ 62.9k · Tools 2 Compare with this →

NPM Sentinel MCP Server 79 · B

AI-powered NPM package analysis MCP server

★ 18 · Tools 19 Compare with this →

MartinLoop MCP Server 79 · B

An independent control layer that gives AI coding agents budgets, brakes, and receipts

★ 410 Compare with this →

GitHub MCP Server 76 · B

GitHub's official MCP server for managing repos, issues, PRs, and workflows via natural language

★ 33.5k · Tools 15 Compare with this →

Compare with other MCP servers

More servers from this repository

Source revision 57e1980e8375 Data synced 2026-10-12 Read the FMRS scoring method