This is an early release preview. You may encounter bugs.
DeepAudit logo
Unclaimed

Tool security

DeepAudit

DeepAudit:人人拥有的 AI 黑客战队,让漏洞挖掘触手可及。国内首个开源的代码漏洞挖掘多智能体系统。小白一键部署运行,自主协作审计 + 自动化沙箱 PoC 验证。支持 Ollama 私有部署 ,一键生成报告。支持中转站。​让安全不再昂贵,让审计不再复杂。

B 77/100 GitHub score ? This grade is derived from GitHub signals, not user votes. Open for the full breakdown.
No votes yet

01 / About

What DeepAudit is.

A code audit platform driven by large language models. It reviews source code across five dimensions -- potential bugs, security vulnerabilities, performance bottlenecks, code style, and maintainability -- and reports each finding in a What-Why-How structure: what the problem is, why it matters, and a concrete code example of how to fix it, with the exact line and column.

Projects can be imported from GitHub or GitLab with a chosen branch and exclusion patterns, or code can be pasted directly into the instant-analysis page for a quick result. Audit tasks report progress as they run, and the results roll up into a dashboard with a 0-100 quality score, problem counts by type and severity, and quality trends over time. Reports export as JSON for downstream processing or as PDF through the browser print dialog.

Model access is configurable at runtime from an admin page rather than at build time, so keys and parameters can change without rebuilding the image. Requests can be routed through a custom base URL for a proxy or self-hosted endpoint, and custom HTTP headers are supported; three API shapes are handled -- OpenAI-compatible, Gemini, and Claude. Analysis behaviour is tunable through concurrency, request interval, timeout, temperature, maximum output tokens, and a per-run file limit, and output language switches between Chinese and English.

Storage has three modes: a browser-local IndexedDB store that needs no configuration, a Supabase cloud database for multi-device sync, and a demo mode that persists nothing. Local data can be exported to JSON and imported back. The front end is React, TypeScript, and Vite with Tailwind CSS and Radix UI, and an in-app log view records user actions, failed API requests, and uncaught exceptions, with filtering and JSON or CSV export.

Features

  • Five-dimension review: bugs, security issues, performance, style, and maintainability in one pass
  • What-Why-How findings: each issue states the problem, its risk, and a usable fix, located to line and column
  • Repository import: GitHub and GitLab projects across JavaScript, TypeScript, Python, Java, Go, Rust, and other languages, with branch selection
  • Instant analysis: paste a snippet and get a score and suggestions without creating a project
  • Runtime configuration: model provider, keys, endpoints, and analysis parameters set in the browser and stored locally
  • Provider choice: Gemini, OpenAI, Claude, Qwen, DeepSeek, Zhipu GLM, Kimi, ERNIE, MiniMax, Doubao, and local Ollama models
  • Proxy support: custom base URLs and headers for relay services and self-hosted endpoints
  • Storage modes: local IndexedDB, Supabase cloud sync, or non-persistent demo mode, with JSON export and import
  • Reports: dashboard scoring, severity breakdowns, quality trends, and JSON or PDF export
  • Deployment: Docker image, Docker Compose, Vercel, or a local Node development server

02 / Discussion CREDIBILITY-GATED

Discussion

Reading is open to everyone. Posting and voting need a verified identity or a GitHub grade of B or higher.

  • No discussions yet.

03 / Build

Build with DeepAudit.

Browse the catalogue for frameworks, tools, and harnesses, each scored on real GitHub credibility.

Get DeepAudit →

Browse the catalogue