chore: wire the Bash tool into the benchmark's PreToolUse matcher - #66
Merged
Merged
Conversation
algomaster99
changed the title
algomaster99
changed the base branch from
fix/bash-manifest-write-detection
to
main
September 25, 2026 11:20
run_case.sh generated a `hook`-condition settings.json with the matcher hardcoded to "Write|Edit", so the Bash-write-detection heuristic added in a prior commit was never exercised by any benchmark run - Claude Code never routed Bash tool calls to the hook at all, regardless of whether they touched a manifest. Every recorded run's "0 Bash interceptions" reflected a wiring gap, not evidence the feature works or that no bypass attempts occurred. Depends on the fix in #65 for the underlying Bash heuristic's false positives; landing this before that fix would flood benchmark transcripts with spurious blocks on read-only commands. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
algomaster99
force-pushed
the
feat/benchmark-bash-hook-matcher
branch
from
September 25, 2026 11:21
0936c4a to
294f558
Compare
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Summary
benchmark/run_case.shgenerates ahook-conditionsettings.jsonwith thePreToolUsematcher hardcoded to"Write|Edit", so the Bash-write-detection heuristic (added for the manifest-bypass-via-bash case) was never actually exercised by any benchmark run — Claude Code doesn't route Bash tool calls to the hook at all when the matcher excludes them, regardless of what those commands touch. Any past benchmark result showing zero Bash interceptions reflects this wiring gap rather than the feature working, or no bypass attempts happening.Change
run_case.sh: matcher for thehookcondition is now"Write|Edit|Bash".🤖 Generated with Claude Code