docs: update evaluation and compaction documentation#3044
Merged
Conversation
The eval harness copies the docker-agent binary from docker/docker-agent:edge into custom base images at build time and overrides their entrypoint with its own /run.sh wrapper. Users need to know their base image's entrypoint will be replaced and that the base image should only provide the runtime environment. Ref: #3029
…windows After the fix in #3042, the summary and keep-tail token budgets used during session compaction scale proportionally to provider_opts.context_size instead of using absolute 16k/20k constants. Small-context-window models (≤ ~16k) no longer have their history wiped during compaction. Ref: #3042
docker-agent
left a comment
docker-agent
left a comment
There was a problem hiding this comment.
Assessment: 🟢 APPROVE
Documentation-only PR. All factual claims were verified against the source code:
docs/features/evaluation/index.md— The new### Custom Base Imagessection accurately describes the eval harness behavior: the binary is indeed copied fromdocker/docker-agent:edgein the Dockerfile template, the entrypoint is correctly overridden unconditionally, and the anchor link#custom-base-imagesresolves correctly.docs/providers/dmr/index.md— The compaction scaling claim is accurate:summaryTokenBudget = min(16000, contextLimit/4)andkeepTokenBudget = min(20000, contextLimit/5)both scale proportionally withcontext_size, confirming small context windows (e.g. 8k) are handled correctly.
No bugs or inaccuracies found in the added documentation.
aheritier
added
area/agent
dgageot
approved these changes
Jun 10, 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
3 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.
Documentation updates
This PR updates documentation to reflect two recently merged code changes.
Changes
--base-imageeval flagcontext_sizefor small windowsDetails
docs/features/evaluation/index.md— Added explanation of how the eval harness handles custom base images: the docker-agent binary is injected fromdocker/docker-agent:edgeat build time and the base image's entrypoint is overridden. Users should provide only the runtime environment in their base image.docs/providers/dmr/index.md(or troubleshooting doc) — Added note that auto-compaction scales summary and keep-tail budgets proportionally toprovider_opts.context_size, so small context windows (e.g. 8k local models) no longer lose session history during compaction.PRs reviewed and found up to date
allow_private_ipssemantics unchanged--session-read-only)