A URI Scheme for AI Agent Prompts

Gratuitous AI related Image of 
Max Headroom by juzmental on DeviantArt

So, I made some minor editorial changes, built out an MCP to demonstrate its use, and submitted my draft The prompt URI Scheme for AI Agent Sessions and Prompts to the IETF.  We'll see where this goes, it's my first time through this process.

At the same time, if you look at AudaciousInquiry/prompt-url: The prompt: URL scheme for AI agent session provenance project on GitHub, you'll see a readily installable MCP and skill for using it in that repo.  The skill documents one use of the MCP, to create a URL for annotating commits, and another use enabling resolution of the prompt via the MCP so that the prompt truly is a traceable link on
the machine and user account where it was run from.

There should be a second skill that describes the use of the MCP to actually find the right Prompt-URL: value to use for a github checkin.  Both the agents I use already know how to do this so I should have one of them write it up.  By the time you see this, it might already be present in the .claude folder.

For what it's worth, GitHub Copilot reads CLAUDE.md and .claude/skills folders so you only have to write things once for Claude and Copilot will use the tooling.  

The MCP framework is highly reusable, and eventually I will extract it from the base MCP APIs.  It runs in node.js in three modes, as a standalone HTTP server, in stdio mode launching a server if one is not already running and connecting to it over HTTP when commands are sent to it, or via CLI where it basically does the same kind of thing.  This makes standalone MCP testing very easy, because the framework handles transport and the rest, so you can just implement commands and get three different ways to activate them.

     Keith

A URI Scheme for AI Agent Prompts

If you've ever read John Moehrke's blog you know he talks a lot about provenance.  One of the challenges I've been faced with recently is the need to track the provenance of an AI output.  The things I need to know are:

1. What was the prompt that produced this, so I can reproduce it again at a later date (this can be used to detect drift and change).
2. What files or inputs were used to produce it.  These should be URIs where feasible so that I can get to the content, not just summary descriptions found in the prompt, and when tied to an HTTP address, even the date of retrieval for the content used.
3. What agent generated it.  Like many, I'm using multiple agents.  What version of the agent is also useful.
4. What LLM and version were used.  Many agents support multiple models.

The first of these is actually not too difficult to identify.  I wanted this to be a URI as well, but right now there is no URI for a prompt, so I made up a syntax.  It follows and can also be found here.  I'm interested in getting community feedback before I spend any time trying to submit this to IETF.