为 AI 代理提供页面自动化和直接 HTTP 控制,用于授权的网络红队。配置沙箱和网络限制,然后审查会话。纯 Rust,无 Chromium 或 V8。
AI 翻译
Give AI agents page automation and direct HTTP control for authorized web red teaming. Configure sandbox and network limits, then review the session. Pure Rust, no Chromium or V8.
安全Rust · LeanGitHub @h5i-dev在线运营者未确认
试用过了吗? 点一下就好,无需登录。
快速、用于 AI 代理的红队浏览器,具备直接 HTTP 流量控制、沙箱执行和可审计会话。纯 Rust 编写。
AI 翻译
Fast, red-teaming browser built for AI agents, with direct HTTP traffic control, sandboxed execution, and auditable sessions. Pure Rust.
取自仓库 README
正在加载完整 README。 以下为已保存的摘录。
h5i (pronounced high-five) is a lightweight browser built for AI agents to test web applications. It combines page automation with direct control over HTTP traffic, letting agents capture, inspect, modify, replay, and compare requests through one interface. Sandboxing, network policies, credential isolation, and auditable sessions keep autonomous testing contained and within scope.
Agent-native red teaming
Browser + direct HTTP control
Attack replay in CI
Replay attack flows in CI
Sandboxed & auditable
Isolated execution with logs
Fast & lightweight
~3× faster, ~86% less memory查看完整 README ↗h5i (pronounced high-five) is an Axum-based Rust web framework designed for formal verification with Lean 4 and optional GPU acceleration. It also provides integrated development infrastructure for browser automation, red-teaming, security regression testing, and sandboxed AI agents.
# Add the Rust web framework
cargo add h5i-app
# Install the h5i CLI and development tools
curl -fsSL https://h5i.dev/install.sh | sh -s -- --websec --recon --test
# AI agent integration (optional)
npx skills add h5i-dev/h5i
h5i-app is a composable, Axum-based Rust web framework built around a verifiable application kernel.
[dependencies]
h5i-app = { version = "0.5.3", features = ["http", "postgres"] }
h5i provides a unified CLI for building, verifying, debugging, and security-testing web applications with AI agents. For example, you can run professional red-teaming simply by opening your favorite AI agent harness and saying "Audit https://<your-target>.com with h5i."
The red-team, CI/CD, sandbox, and UI are framework-independent, so there's no need to adopt
h5i-app.
CLI Examples
# Project tooling: Scaffold Rust web applications, extract Lean 4 models, check proofs, and test proof sensitivity.
h5i app new counter && cd counter
h5i app extract
h5i app prove
h5i app mutate
# Browser automation: Navigate websites, interact with forms, and inspect authenticated workflows.
h5i browser open https://example.com
h5i browser click @e3
# HTTP security testing: Capture, modify, replay, and compare HTTP requests.
h5i websec requests
h5i websec replay req_42 --set query.id=456
# Reconnaissance: Discover endpoints, parameters, and hidden attack surfaces.
h5i recon crawl --max-requests 200
# CI/CD: Replay security tests and enforce API coverage requirements.
h5i test .h5i-tests/tests/auth.yaml --target https://staging.example.com
h5i test --target https://staging.example.com --openapi openapi.yaml --min-coverage 80
# Sandboxed environment: Run isolated workflows with filesystem and network restrictions.
h5i box --name lab --profile browser
h5i box shell lab -- claude
# Dashboard: Monitor browser sessions and agent activity through a local web interface.
h5i ui
h5i is licensed under the Apache License 2.0. See LICENSE.