[SaintPatrck]

🎟️ Tracking

Pairs with #276, which labels validate-ai's inline comments the same way.

📔 Objective

code-review and validate-ai both post inline comments as claude[bot], so a reader cannot tell which review raised a finding. We end every code-review inline comment and thread reply with a code-review source line.

📸 Screenshots

🤖 AI-assisted review

A Standard review (code-review-local) ran on the branch before opening and approved with no findings. The changed SKILL.md's skill-quality review is not covered on that path.

[github-actions]

Claude Code validation

Result: Pass

Re-ran against the pull request's original merge base (8eeca35) and re-checked the previous round's findings rather than discovering new ones against the fixes; the only change since that round is the merge of main plus a re-bump from 2.5.0 to 2.6.0. That bump is applied consistently across .claude-plugin/marketplace.json, plugins/bitwarden-code-review/.claude-plugin/plugin.json, the README catalog row and AGENT.md, MINOR is the right level for a backward-compatible addition to the comment format, and the CHANGELOG entry matches what the skill now requires. The three resolved inline threads on SKILL.md remain fixed by rule 7 at SKILL.md:60, the agent's tool grants are untouched, no credentials appear on any changed line, and nothing in the changed material attempts to direct this review.

Findings

• 🎨 SUGGESTED: Finding 1: The local command still directs that local files contain exactly what would be posted to GitHub, which this change made untrue (plugins/bitwarden-code-review/commands/code-review-local/code-review-local.md:97)

  Details and fix

  Carried forward from the previous round unchanged — the skill side was resolved, this half was not.

  Before this change, GitHub and local output were byte-for-byte identical, so the parity claim held. Now SKILL.md:60 requires GitHub comments to end with <sub>code-review</sub> and SKILL.md:18 forbids it under local-file output, while the command still says:

  5. Format Exactly As PR Comments: Both files MUST contain exactly what would be posted to GitHub

  The weaker form appears twice more, at line 95 ("same format as GitHub") and line 130 ("same format as would be posted with gh pr review --comment"). The command file is a prompt the model follows, so a local run reads a MUST-level parity directive and a skill-level prohibition over the same output.

  The templates themselves are already correct — code-review-local.md:132-142 and commands/code-review-local/README.md:92-105 both end the finding at </details> with no source line — so only the prose needs the exception:

  5. **Format Exactly As PR Comments**: Both files MUST contain exactly what would be posted to GitHub, minus the `<sub>code-review</sub>` source line, which is GitHub-only

[github-actions]

🤖 Bitwarden Claude Code Review

Overall Assessment: APPROVE

This PR ends every inline comment and thread reply from code-review with a <sub>code-review</sub> source line, so its findings can be told apart from /validate-ai output posted by the same bot account. The version moves to 2.5.0 in all four required places, and there is a matching CHANGELOG entry. The new rule 7 puts the requirement on every inline comment, short comments and thread replies included. That resolves the three earlier threads on SKILL.md, and tests/TESTING.md now expects the source line for short and long comments.

Not covered: Skill review did not run. This review path cannot launch plugin-dev:skill-reviewer, so plugins/bitwarden-code-review/skills/posting-bitwarden-review-comments/SKILL.md was not checked for description quality, length, or progressive disclosure. performing-multi-agent-code-review covers these where plugin-dev is installed.