Repository navigation
Keep critical task evidence together and gate incomplete decisions - #90
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Task retrieval could retain a chosen approach while dropping a weakly matched constraint, adjacent exception or rejected alternative. Complete evidence-bound families and relationships are now selected and budgeted together, and their qualification/history facts survive optional decision prose packing.
Incomplete retrieval (including singleton budget omissions and truncated queries/indexes) returns explicit incomplete context instead of decision readiness or the exact-reference shortcut. Concept expansion requires current documentary support and matching scope; existing candidate and relation-hop bounds stay in place.
The checked-in manifest pins source hashes, critical record IDs, exact evidence IDs and budgets for all five requested cases and two negative controls. Tests exercise the actual
--fastCLI with zero model calls, whole-group omission, semantic authority controls and packed responses. Retrieval and runtime bounds are documented indocs/DECISION_LENSES.md.Validation:
cargo fmt --all -- --checkcargo clippy --all-targets --locked -- -D warningscargo test --all-targets --locked(429 tests passed; host access for localhost test listeners)python3.14 -m unittest discover -s evaluation/tests -v(271 tests, one existing skip)No live model evaluation was run. Tight budgets may now return incomplete context rather than a recommendation.
Closes #82.