
Cross-Modal Review
Gets a second, different AI model to check your work before you commit it
- Repository
- garrytan/gbrain
- Stars
- 31,447
- Licence
- MIT
- Files
- 1 file
What it does
When installed, the agent sends its work (a code diff, a brain page, an analysis or a decision) to a different AI model for an independent review, graded against the original promise or Contract for that work. It reports where the two models agree and disagree, and never applies the reviewer's suggestions without your explicit approval. If one model refuses, it quietly tries the next model in a chain, and it can hand code diffs to an external Codex review tool.
Great for
- Developers who want a second opinion on significant code changes before merging
- Teams making security-sensitive changes such as auth flows or webhook handling
- People running bulk operations or migrations who want a check first
Skip it if
- Simple typo fixes, formatting changes or routine memory writes, which the skill says it should not review (wasteful and noisy)
- Setups without access to a second model provider, since the review cannot run and must not be reported as passed
- Anyone expecting a guaranteed-correct verdict: cross-model agreement is a signal, not proof
Try asking
- “Give me a second opinion on this refactor before I commit it”
- “Challenge this authentication change for security holes using an adversarial review”
- “Double check this migration script before we run it on the full database”
- “Get another perspective on this architecture decision”
Install in 3 steps
- 1
Download the skill
The ZIP contains the 1 original files, the MIT licence and an attribution note.
cross-modal-review.zip - 2
Add it to your agent
Claude Code (all projects)
mkdir -p ~/.claude/skills && unzip cross-modal-review.zip -d ~/.claude/skills/Claude Code / Agent SDK (this project only)
mkdir -p .claude/skills && unzip cross-modal-review.zip -d .claude/skills/Claude apps (web/desktop): Settings → Capabilities → Skills → Upload skill, and choose the ZIP.
- 3
Ask for the task
No special command needed — the agent reads the skill's description and loads it when your request matches. Try one of the prompts above.
Get better results
- Make sure the originating skill's Contract is written down first, since reviews are graded against it
- Configure at least two provider keys so refusal routing has a fallback model
- Review the reviewer's findings yourself; the skill is designed so nothing is applied automatically
What's inside
- SKILL.md
Redistributed unmodified from garrytan/gbrain under the MIT licence. The assessment was drafted with AI from the skill's files on 2026-10-29. Skills can instruct your agent to run code — read SKILL.md and any scripts before installing.