Review Bugbot

Review Bugbot

Use this skill when the user asks to run /review-bugbot.

Launch exactly one bugbot subagent with:

• run_in_background: false unless explicitly asked to run in background
• description: "Bugbot"
• subagent_type: "bugbot"

The review subagent computes the local diff from the repository path, so do not compute the diff yourself before launching it. The repository path should be the active workspace or repository root for the code the user wants reviewed.

By default, the review subagent infers the repository's actual base branch, such as main, when computing branch changes. In most cases, do not provide Base Branch. Only provide it if you know the current branch or PR should be compared against a specific branch other than the repository's default base branch, such as when you created the current branch from another branch.

Special case: if the user explicitly asks to review a specific PR or branch, make sure that target is checked out before launching the subagent:

• Examples include github.com/... /review, review {link}, or review {branch-name}.
• Resolve the provided PR link, PR number, or branch name to the PR head branch or named branch.
• Check whether the target branch is already the currently checked out local branch. If it is, continue.
• If a different branch is currently checked out, try to switch to the target branch.
• If Git refuses to switch because local files would be overwritten, conflicts need resolution, or another checkout blocker occurs, explain the blocker and ask whether the user wants to stash local changes before retrying.
• Only stash after the user confirms. If the stash succeeds, retry switching to the target branch.
• Launch the review subagent only after the target branch is checked out locally.

Use this exact prompt shape:

Full Repository Path: <absolute repository path>
Diff: <one of: "branch changes", "uncommitted changes", "natural language">
Base Branch: <only include this line when reviewing branch changes against a known specific base branch>
Change Description: <required only when Diff is "natural language"; list each changed file and what changed in it>
Custom Instructions: <only include this line when the user gave specific review instructions>

Default to branch changes, which reviews branch changes against the merge-base with the default/base branch, including committed, staged, and unstaged changes. If the user asks to review only uncommitted, local working tree, dirty, or not-yet-committed changes, use uncommitted changes.

If the review subagent fails before producing findings, inspect the failure text.

• If the failure is caused by calling the subagent incorrectly, such as a missing Full Repository Path, missing Diff, wrong prompt shape, or wrong subagent type, correct the invocation and retry it once immediately.

• If the subagent reports it could not compute the diff (for example an empty diff, missing diff metadata, or that it could not compute the diff), retry once with Diff: natural language, omitting Base Branch and providing a Change Description. The subagent will read those files directly. Only use this as a last resort, after the regular diff-based review has failed because the diff could not be computed.

  Write the Change Description as one block per changed file: a <path> (added|modified|deleted|renamed) header followed by bullet points of what changed. Mention line numbers or ranges inline where they help and you know them (for example (L40-58) or around L120); they are optional, not required on every bullet. Example:

  src/auth/login.ts (modified):
  - rewrote validateSession (L40-58) to check token expiry before the DB lookup
  - removed the fallback that accepted empty tokens
  
  src/auth/legacy.ts (deleted)
  
  src/auth/mfa.ts (added):
  - new verifyMfaCode() (L1-30) that calls the TOTP service and rate-limits attempts
  

• For any other subagent failure, retry once with the same prompt shape.

• If the same failure persists after the retry, stop. Briefly tell the user that the review subagent could not complete and include the short error or blocker. Do not keep retrying.

After the subagent finishes, summarize the result:

• If it reports that no diff was found or the diff is empty, tell the user in one sentence that there was no diff to review.
• If it found no issues, give a one-line status such as "Bugbot found 2 findings / Bugbot found no bugs".
• If it found issues, print a compact markdown table with one row per finding, sorted by severity (highest first), using exactly these columns: Severity, Location (file:line), Finding. Put the file and line in the location column as file:line.

Do not fix findings or rerun review unless the user explicitly asks for that next step.