Postman is already Postman for MCP. We launched MCP support several weeks ago, both in generating MCP servers from the public APIs on our network (over 100k) and with the MCP client which can test, debug, and validate MCP servers with full support for streamable http, sse, and stdio and capabilities (tools, prompts, resources).
My platform goes beyond being an automatic wrapper of an API and lets you specify to the model, in natural language, how it should parse inputs and outputs. I find LLMs are very responsive to this type of specification, and to the best of my knowledge no one is trying this yet.
You also don’t seem to offer a simple chat-based client.
notpushkin · 3h ago
Please note that posting the same thing multiple times in succession is frowned upon here (but congrats getting to the first page!)
I’m also confused by the title – why “Postman”? I do know about Postman the HTTP client, but I don’t get the parallel here.
phildenhoff · 3h ago
andes314, can you expand on how you see this as Postman for MCP?
andes314 · 1h ago
It is a platform to create MCP servers from API endpoints, and then chat with them without having to use Claude’s clunky integration process. It is simple and complete.
andes314 · 4h ago
If it enters hug of death, please try Claude client to test it! Anthropic allows me only 30req/s at the moment
fcarraldo · 2h ago
How much are you spending on tokens right now?
andes314 · 1h ago
Less than you’d think
TZubiri · 4h ago
"error, please clear and try again: Error code: 429 - {'type': 'error', 'error': {'type': 'rate_limit_error', 'message': 'This request would exceed the rate limit for your organization (67777b05-661c-4183-aa19-ec6e299f95ac) of 50,000 input tokens per minute. }}
This is a very bad idea buddy. Maybe try letting users set their API tokens.
Check it out! https://postman.com/downloads
as well as our MCP catalog
https://getmcp.dev
You also don’t seem to offer a simple chat-based client.
I’m also confused by the title – why “Postman”? I do know about Postman the HTTP client, but I don’t get the parallel here.
This is a very bad idea buddy. Maybe try letting users set their API tokens.