Merged
Conversation
Signed-off-by: Derek Misler <derek.misler@docker.com>
derekmisler
force-pushed
the
limit-token-usage
branch
from
e7771a8 to
45dc136
Compare
![docker-agent[bot]](https://avatars.githubusercontent.com/in/2787757?s=60&v=4){kind=small}
Contributor
There was a problem hiding this comment.
Assessment: 🟢 APPROVE
This PR successfully reduces token usage by condensing verbose multi-paragraph instructions into tighter prose and extracting the large jq-based GitHub posting template into a dedicated reference file (posting-format.md). The refactoring maintains all operational details while making the agent prompt more efficient.
Changes reviewed:
- ✅ Condensed instructions in
pr-review.yaml(removed redundant explanations) - ✅ Extracted posting format template to
review-pr/agents/refs/posting-format.md - ✅ Added step in
action.ymlto copy reference file to/tmp/posting_format.md - ✅ Removed auto-merge labels from release workflow (unrelated cleanup)
- ✅ Added
--method GETtogh apicall (explicit method specification)
No bugs found in the added code. All changes are documentation/configuration refactoring with no impact on runtime behavior.
krissetto
approved these changes
Apr 7, 2026
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
2 participants
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Related Issues
Closes: https://github.com/docker/gordon/issues/363
Summary
Reduces token usage in the
review-praction by condensing verbose multi-paragraph instructions inpr-review.yamlinto tighter prose, and by extracting the largejq-based GitHub posting template out of the agent prompt and into a dedicated reference file (posting-format.md) that is copied to/tmpat runtime. This keeps the agent prompt lean without losing any operational detail.Token Reduction Estimate
Instruction text: ~6,606 → ~5,334 tokens per call (~19% reduction)
The system prompt is resent on every API roundtrip (tool calls, delegations). Based on a real review run (pinata, single-file PR), the pipeline made ~22 tool-call roundtrips:
~15K input tokens saved per review on a small PR. Larger PRs with multiple chunks and findings would see 30-40 roundtrips, pushing savings to ~20-25K tokens per review.
Tip
Comment
/reviewto trigger the PR Reviewer agent for automated feedback.Comment
/describeto generate a PR description.