Compress visual-explainer prompts - #61
Merged
Merged
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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
This trims the visual-explainer skill and command prompts from 10,824 words to 2,131 words, which is a little over 80% less prompt text loaded by agents.
The idea came from Matt Pocock’s thread on no-op skill instructions: https://x.com/mattpocockuk/status/2069784839474032896
I kept the parts that actually steer behavior: when to render HTML, which references to read, Mermaid shell/zoom rules, table handling, slide navigation, diff review file maps, and the final quality checks. The removed text was mostly repeated explanation and long-form rationale that can live in the reference docs instead of the always-loaded skill and command prompts.