Repository navigation
Add source-pinned real-project decision regression corpus - #94
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Closes #86.
Adds seven documentary decision/application controls across Python PEPs and Go's compatibility policy. The manifest binds immutable upstream revisions, original licensed files, hashes, exact line citations, status witnesses, scope, required conditions, negative cases and explicit unknowns.
The offline tests use existing source/evidence storage and the real
context --fast,decisions,cases,evidenceandauditcommands. They check selection, lifecycle/scope preservation, complete excerpts, budget omissions, supersession grouping, zero model calls and unchanged registry bytes. Both existing evaluators recognize the public snapshots as development fixtures, including after task relabeling. No production retrieval or inference behavior changes.Source labels are implementation-agent annotations; separate human review remains pending, with notes/disagreements retained in the manifest. Case reports record exclusions and unmeasured before/after comparisons. No matching code/OpenWiki snapshot or live-model exploration was curated, and these controls do not establish independent task success or productivity gains in #2.
Validation:
cargo fmt --all -- --checkcargo clippy --all-targets --locked -- -D warningscargo test --all-targets --locked: 444 passed, one live Ollama test ignored (host execution for localhost test servers)