fix(bitwarden-code-review): grant git status where a local review is told to run it - #220
Conversation
…told to run it `code-review-local` instructs the reviewer to inspect the working tree with `git status`, and its README documents that behavior, but neither the command nor the agent listed the grant — so the call prompts or fails on the local path while the multi-agent path, which already grants it, works.
Claude Code validationResult: Pass Validated PR #220 against The new tool grant is read-only, prefix-scoped in the same shape as the existing CriticalNone. MajorNone. Minor
Pre-existing (not counted against this changeset)
What was validated
Checks run
|
🤖 Bitwarden Claude Code ReviewOverall Assessment: APPROVE This adds No findings. |
🎟️ Tracking
Layer 1 of 4 in stack #221, bottom to top: #220, #217, #219, #218. The first three each bump
bitwarden-code-review, so they merge in that order.No Jira ticket. Noticed while auditing tool grants in
bitwarden-code-review.📔 Objective
code-review-localtells the reviewer to inspect the working tree withgit diffandgit status, and its README documents the same behavior. Neither the command'sallowed-toolsnor the agent'stoolslistedgit status, so the call prompts or fails on the local path.performing-multi-agent-code-reviewalready grants it, which is why the multi-agent path works and this one does not.Adds
Bash(git status:*)to both.One thing this does not fix: the agent's local mode fetches only committed work, while the command promises "uncommitted and committed changes." That is a behavior change rather than a grant fix, so it is left alone here.
Version 1.14.0 to 1.14.1.