Benchline

Free, open-source developer tools for you and your AI agent: JSON, PDF, images, JWT, DNS and llms.txt in the browser, plus one MCP URL your agent can call.

Comment

Inspiration

What it does

How I built it

Challenges I ran into

Accomplishments that I'm proud of

What I learned

What's next for Benchline

Inspiration

I kept switching between a dozen utility sites for JSON, PDFs and tokens, and my coding agent couldn't use any of them.

What it does

Benchline is an MIT-licensed set of developer tools: JSON/YAML/TOML/CSV conversion and formatting, PDF merge and compress, image conversion and OCR, JWT decoding, hashing, DNS, TLS and HTTP header checks, and llms.txt generation. Most of them run fully client-side, so your files never leave the browser.

Every tool is also exposed on one stateless MCP server. Add https://benchline.fabwaseem.com/mcp to Cursor, Claude, VS Code or any MCP client and your agent gets the whole toolbox, with no API key, no account and no session. The same functions also have a plain HTTP API, and it's self-hostable with Docker.

How I built it

Next.js and TypeScript. One tool registry generates the UI, the MCP tools and the HTTP routes, so adding a tool takes one registry entry. The MCP server is stateless Streamable HTTP under the 2026-07-28 spec.

Challenges I ran into

Keeping numbers above 2^53 exact when formatting JSON (fixed after community feedback), and adding SSRF protection to the network tools.

Accomplishments that I'm proud of

It's listed in the official MCP Registry and on Glama, and the same code serves people in the browser and agents over MCP.

What I learned

Designing tools once for both humans and agents forces clean inputs and outputs, and stateless MCP makes hosting simple.

What's next for Benchline

More tools and a Docker image on Docker Hub.

Built With

Share this project:

Updates