mcpgen tool auto-converts OpenAPI specs into ready-to-use MCP servers

·1 views

A developer has released mcpgen, an open-source Python tool that automatically generates Model Context Protocol (MCP) servers from OpenAPI 3.0 and 3.1 JSON or YAML specification files. The tool creates one MCP tool per API endpoint, handles authentication methods such as API keys and Bearer tokens, and outputs a readable Python script with minimal dependencies. Users can install it via pip and generate a working server in a single command, which can then be connected to clients like Claude Desktop. The project was validated against a real-world OpenAPI 3.1 fixture involving tweet and user endpoints, passing all 18 tests on the first attempt. Available on GitHub and PyPI under the MIT license, mcpgen aims to eliminate boilerplate code for developers building LLM agents that interact with REST APIs.

Read the full story at DEV Community

This is an AI-generated summary. ShortSingh links to the original source for the complete article.

Discussion (0)

SKILLmama v1.3 Scans Your Project and Flags Missing Tech Before You Ask

Developer tool SKILLmama has been updated to version 1.3 with a new proactive workflow that scans a project's files and identifies capability gaps without requiring a specific query from the user. The tool reads package files, config files, infrastructure definitions, and source structure to build a Stack Profile listing detected and missing components such as auth, caching, observability, and queuing. It then ranks the gaps by severity — high, medium, or low — based on what is typical for the detected project type. Before proceeding to search and recommend libraries, SKILLmama pauses to ask the user three clarifying questions and waits for a response. The original command-based workflow, where users specify a capability they need, remains available alongside the new scan-first flow.

0 comments Read more at DEV Community

ProgrammingDEV Community ·

Developer Builds Custom Guitalele Notation App After Finding No Tools Exist

A developer began building a web-based music notation tool roughly three weeks ago after finding almost no digital resources — such as tabs, tuners, or scores — available for the guitalele, a niche string instrument. What started as a simple text area to parse personal musical shorthand quickly expanded into a full editor with metadata fields, score management, publish/draft toggles, and an auto-resizing input, all stored locally using React state. The developer invented their own shorthand notation system, such as '3:1@q' to represent fret, string, and duration, and wrote custom parser functions to handle notes, chords, rests, and ties. However, after testing the first working version by entering an original tab, the notation proved difficult to type and confusing to read in practice. The project is ongoing, with the developer identifying missing features like two-voice polyphony and measure validation as the next challenges to solve.

0 comments Read more at DEV Community

ProgrammingDEV Community ·

MiMo V2.5 Pro Outperforms DeepSeek V4 Pro at Debugging but Loses on Speed

A developer on DEV Community ran a real-world debugging test pitting DeepSeek V4 Pro against MiMo V2.5 Pro using a genuine race condition bug from the open-source httpcore library. Both models were given the full project codebase before the official fix and asked to identify the root cause and propose a solution. MiMo found three race conditions versus DeepSeek's one, delivered deeper analysis, and cost slightly less at $0.13 compared to $0.14. However, DeepSeek completed the task in roughly eight minutes while MiMo took about fifteen, using fewer tokens overall. The comparison suggests MiMo has an edge for debugging tasks, while DeepSeek may be better suited for faster code-writing scenarios.

0 comments Read more at DEV Community

ProgrammingDEV Community ·

Developer Tests Claude Code and GitHub Copilot for 30 Days, Finds Each Has Clear Strengths

A software developer spent 30 days using Claude Code exclusively before returning to GitHub Copilot for a week, testing both tools across real projects including a React dashboard, a Python data pipeline, and legacy codebases. GitHub Copilot was found superior for fast, inline code completion, excelling when the developer already knew what to write and needed to type quickly. Claude Code outperformed in tasks requiring broader codebase understanding, such as explaining authentication flows across 50,000-line codebases, generating full feature implementations from descriptions, and setting up CI/CD pipelines autonomously. Claude Code also proved more effective for debugging by reasoning over symptoms and for learning unfamiliar technologies through conversational back-and-forth. The developer concluded that the two tools serve fundamentally different purposes rather than being direct competitors.

0 comments Read more at DEV Community