Exposing Hypermedia Information Graphs to LLMs #359
+548
−0
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
- Restore Korzybski quote and trail race metaphor - Restore extended Claude thinking log evidence - Restore curiosity-driven evaluation insight - Restore guidance architecture intro paragraph - Expand summary paragraph with key techniques - Add llms.txt comparison (Stripe, Claude Code) - Align examples to CVE/security domain - Merge vector search section into Background 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Show that workflows are formal enough for mechanical translation: - Add `list` command output showing enumerable workflows - Add generated bash script demonstrating precise translation - Script shows curl/jq-based HAL navigation with error handling 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Rendered
We are in the process of replacing our existing
releases-index.jsondocument with a hypermedia graph, per #358. This spec defines an approach and testing methodology for efficiently exposing the graph to LLMs. The results are positive and illuminating. It turns out that initial design goals of the graph per already well-aligned with the LLM cost model. Additional opportunities for alignment were discovered through the testing process and became part of the proposed model.This is the intended user experience:
Note: This works with claude.ai, ChatGPT, and direct API use. Works best with claude.ai. You can try that actual prompt or one like it.
Prototype LLM entrypoints for graph:
llms.txtis to some degree the standard model as an LLM on-ramp to a custom system. A JSON variant was developed as an alternative to requiring markdown. The addition of a second entrypoint isn't competition between the formats, but an interest in motivating an LLM to use JSON as both data and guidance router. The spec goes into much more detail on the role ofllms.jsonand how it relates toindex.jsonas the core data entrypoint to the graph.