Allow tools to have an output schema #273
marianogonzalez
started this conversation in
Ideas - General
Replies: 1 comment 2 replies
-
|
I would also love this but why not something like this: Though I would be overjoyed if MCP suddenly decided to just simply support the OpenAPI specs, it seems they wanna do their own thing, but an output schema is absolutely required IMO |
Beta Was this translation helpful? Give feedback.
2 replies
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
-
Pre-submission Checklist
Your Idea
Tools right now are open ended when it comes to the produced output. This makes sense when looked from an LLM point of view.
However, a fair amount of "written by a human" tools will tend to return data in a fixed schema (e.g: data driven tools that will return JSON in a fixed schema). As MCP adoption grows larger in the enterprise, the ability to catalog tools and have a strong input-output contract about them is greatly desirable. For tools used in Agent to Agent communication, this could even be used as a way of hint an LLM to produce data in a certain format and schema, without the need of prompt engineering.
This should of course be an optional feature, since a) this is not always needed, and b) need to preserve backwards compatibility.
Example 1: A tool that transforms an address to coordinates
Example 2: Same tool, but returning the result as an embedded resource instead
Scope
Beta Was this translation helpful? Give feedback.
All reactions