# AI Review > Review Gemini API agent code against mandatory rules in backend/AGENTS.md Use when this capability is needed. - Skill: `tomevault-io/ai-review` (Agent Skill, multi-file: 2 files) - Install (CLI): `npx skillmds@latest add tomevault-io/ai-review` - Raw SKILL.md: https://api.skillmd.com/api/skills/tomevault-io/ai-review/raw - Safety review: pending (external: skill-scanner PASS, skillspector PASS) - Works with: Claude Code, Claude.ai, OpenAI Codex - Category: AI & ML - Author: tomevault-io (https://skillmd.com/u/tomevault-io) - Updated: 2026-09-17 - Page: https://skillmd.com/skills/tomevault-io/ai-review --- ## What I do 1. Read `backend/AGENTS.md` to get the mandatory rules for Gemini API 2. Review the changed or new files that contain Gemini API agent code 3. Check for: - Rate limiting (10/hour per user for AI endpoints) - Proper structured output schema handling - Error handling and retry logic - Correct client initialization - No hardcoded API keys 4. Report any violations with specific line references ## When to use me Use this after creating or modifying any Gemini API agent code. This should be run as part of the code review process. ## Important notes - The rules in `backend/AGENTS.md` are mandatory for all AI agent implementations - Focus on the specific rules in that file - don't make up additional requirements - Provide actionable feedback with file:line references for any issues found --- > Source: [Arkokat/hearthkeeper](https://github.com/Arkokat/hearthkeeper) — distributed by [TomeVault](https://tomevault.io).