Repository navigation
v0.3.1 — aggregation honesty, singleton clusters, report accuracy - #1
Merged
Merged
Conversation
Cross-page fault clusters that group by identical message with no distinguishing stack are now rendered under a separate "Possible duplicates" heading, labelled "confidence: low · message only" with an explicit caution that the shared root cause is NOT proven. Reliable (stack + message) clusters keep the confirmed root-cause grouping. Every cluster still exposes its full per-page member detail (status, stability, repro, evidence link) so raw findings remain traceable. Adds 3 aggregate data-contract tests (stack vs stack-less stay separate; message-only machine-distinguishable; per-member detail preserved) and a real-browser report test asserting the honest two-tier labeling.
The combined issue document printed the first finding's verification attempt count as if it were a global "N independent browser runs" figure, which is wrong (attempts vary per finding). The header now states the verified/suspected meaning generically and points to each finding's own ratio, which already appears in every finding body. Adds unit tests proving: issues.md keeps every original finding (id + evidence + stack) even when some share a cluster; the fabricated global count is gone; and CJK/emoji/markup payloads are escaped in the offline HTML without mangling non-ASCII content.
… > soft) Adds a real-browser regression for the read-only-first click policy proving that exploration.allowSelectors can never revive a HARD-dangerous control (delete/pay stay forbidden) while a soft keyword becomes clickable only when explicitly authorized, and denySelectors exclude a control outright. Closes the one Phase 4 security property that had no direct test.
…ding A one-page message-only signature was being emitted as a "cluster" and, after the honesty change, rendered as a "possible duplicate" of a single page — a mislabel. aggregateFindings now only produces a cluster when a signature spans two or more pages; singletons remain in `findings` (which drives compare and the per-finding report), so no information is lost. Verified against a real demo scan: the spurious singleton cluster is gone. Strengthens the wrong-merge / miss-merge guards to prove non-merging via two-page clusters and adds an explicit "lone fault is not a cluster" test.
Bump 0.3.0 -> 0.3.1 (package.json, lock, runtime VERSION). Add the v0.3.1 CHANGELOG entry describing the cluster-honesty, singleton-cluster and issues.md header fixes. Update the README (en + zh) Status section and the compatibility matrix with the real verified suite size (122 -> 132) and an honest note on which CI matrix run each column reflects.
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.
Reliability/trust refinement on the verified v0.3.0 base (989d137).
Head commit: 8cb45ef