Release 1.17.0: fix retrieval regressions and accelerate duplicate scans - #41
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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 release fixes lost structural metadata in small files, improves bounded context selection and cluster coherence, exposes graph resolution evidence, and accelerates duplicate scans. Search now accepts a serialized JSON byte budget and distinguishes ranking strength from answer sufficiency.
Version core, CLI, and MCP to 1.17.0 and synchronize package dependencies, lockfile, server metadata, and plugin manifests. Existing indexes rebuild under chunking policy 6 on the next explicit index command.
Validation: workspace build/typechecks passed; 232 full-suite tests plus two subsequently added regression tests passed. Real Ollama fixture covered small TypeScript/Python/Java/Vue files and alias call tracing. The same five-second duplicate scan processed 873 candidates versus 44 previously, with continuation equivalent to an uninterrupted run. JSON byte-budget checks passed for all projection modes. Detailed results are in docs/evaluation-fixes-2026-09-24.md.