Automatic Multi-File Code Review Agent
This agent performs automated code review across multiple files simultaneously, analyzing quality, security, performance, and architectural consistency. It produces a structured report with recommendations prioritized by severity, refactoring suggestions, and corrected code snippets ready to integrate.
For who
Developers, tech leads, and development teams looking to automate and systematize their code reviews before merge or deployment.
Input
Le code source de 2 à 10 fichiers à reviewer, collés avec leur nom de fichier en en-tête (ex: // fichier: src/auth.ts). Peut inclure tout langage : JavaScript, TypeScript, Python, Java, Go, etc.
steps (4)
File Inventory and Mapping
promptAnalyzes the structure of submitted files and identifies dependencies between them.
Quality Analysis and Bug Detection
promptExamines each file to detect bugs, code smells, and best practice violations.
Security Audit and Compliance
promptDetects security vulnerabilities and verifies compliance with standards.
Recommendations and Corrected Code
promptSynthesizes all analyses and produces fixes ready to integrate.
Output
Rapport de code review complet en Markdown contenant : cartographie des fichiers, liste des bugs et code smells priorisés, audit de sécurité OWASP, correctifs de code prêts à intégrer, recommandations de refactoring et checklist de validation.
Example
Input
// file: src/controllers/userController.ts
import { db } from '../db';
export async function getUser(req, res) {
const user = await db.query(`SELECT * FROM users WHERE id = ${req.params.id}`);
res.json({ password: user.password, ...user });
}
// file: src/routes/api.ts
import { getUser } from '../controllers/userController';
router.get('/user/:id', getUser);
// file: src/middleware/auth.ts
export function checkAuth(req, res, next) {
if (req.headers.token == 'admin123') next();
else res.status(401).send('Unauthorized');
}Output
## Executive Summary
**Score: 22/100** — Changes required
- 3 critical, 2 major, 1 minor
### Top 3 Urgent Issues
1. **SQL Injection** (userController.ts:3) — Direct concatenation of req.params.id
2. **Password Exposure** (userController.ts:4) — Password hash is returned in the API response
3. **Hardcoded Secret** (auth.ts:2) — Admin token in plain text in source code
### Critical Fix #1 — SQL Injection
```typescript
// BEFORE (vulnerable)
const user = await db.query(`SELECT * FROM users WHERE id = ${req.params.id}`);
// AFTER (secure)
const user = await db.query('SELECT id, email, name FROM users WHERE id = $1', [req.params.id]);
```
[...full report with all fixes, recommendations and checklist]Customization
| Parameter | Description | Default |
|---|---|---|
| review_standard | Standard ou guide de style Ă appliquer lors de la review (ex: Airbnb, Google, conventions internes) | Bonnes pratiques universelles (SOLID, Clean Code, OWASP Top 10) |
| severity_filter | Niveau minimum de sévérité à inclure dans le rapport (critique, majeur, mineur, tous) | tous |
| focus_areas | Domaines d'analyse prioritaires séparés par virgules (sécurité, performance, maintenabilité, tests) | sécurité, qualité, maintenabilité |
Technical Notes
This agent is designed to process between 2 and 10 files simultaneously. For larger codebases, split into coherent batches by module or feature. Paste each file with a header comment indicating its relative path (e.g., // file: src/auth/middleware.ts) to allow the agent to reconstruct the dependency graph.
The security audit step is based on the OWASP Top 10 2021 framework. For specific compliance audits (PCI-DSS, HIPAA, SOC 2), specify the standard in the review_standard parameter. Generated code fixes are directly applicable but should be validated by tests before merging.
For CI/CD usage, you can chain this agent with a unit test generation agent to automatically cover critical cases identified during the review.
Related Prompts
Prompt to Write a Professional Quote Request Email
A complete prompt to generate a structured, professional and customizable quote request email adapted to your industry and recipient.
Managing Imposter Syndrome
Student well-being
Write an Effective Professional Complaint Email
This prompt generates a structured, professional and persuasive complaint email adapted to any type of commercial or service dispute.
Gemini Prompt to Optimize a Workflow
In a professional environment where every minute counts, workflow optimization has become an indispensable strategic lever. Google Gemini, thanks to its contextual analysis capabilities and deep understanding of business processes, positions itself as a powerful ally for rethinking your workflows. Whether you are looking to eliminate bottlenecks in a production chain, automate repetitive tasks in an administrative process, or streamline collaboration between teams, Gemini can analyze your existing workflow, identify inefficiencies, and propose concrete, actionable improvements. Its multimodal approach allows it to process both textual descriptions and process diagrams, offering a 360-degree analysis of your operations. By crafting the right prompt, you transform Gemini into an optimization consultant capable of mapping your processes, quantifying time losses, and suggesting restructuring tailored to your specific context. This guide provides you with the most effective prompts to fully leverage this capability.