← 开源
hardenrun

aif

On device Firewall for Coding and Personal Agents

AI EngineeringTest & guardGive agent tools
在 GitHub 打开
增长势头
+424 小时新增 Star+0.3%
1.52k
Star
392
Fork
+253
本周
1
贡献者
创建于 2026-08-01 · 更新于 2026-10-12 · 今日第 2380 名
主要开发者
README

Harden.Run AIF (Agentic Integrity Foundation)

The Harden AIF is a local system for coding agent safety. If AIF is useful to you please star this repo.

Getting Started

curl -fsSL https://aif.harden.run/install.sh | sh

Then:

aif configure
aif demo
aif show
aif show --tui

harden-aif-synchronous-ltr-16x9-light

Supported agents

Agent Coverage
Claude Code native hooks, pre-execution blocking
Codex native hooks, pre-execution blocking
Cursor (IDE + CLI) hooks, pre-execution blocking
Gemini CLI hooks, pre-execution blocking
Kiro hooks for the main agent; subagent MCP calls via proxy
OpenClaw, Hermes block-and-steer via a bridge process

See more at our trust page.

Telemetry

We never collect code, commands, file paths, prompts, or anything a secret could ride in. We collect anonymous operational counters only: version, OS/arch, adapters in use, decision/block counts.

  • aif telemetry show prints the exact payload before it is sent — a secret-safety tool inspecting its own egress.
  • Opt out with one environment variable: AIF_NO_TELEMETRY=1.

For full schema see trust.

Dashboard

Use aif show to see the local dashboard at http://127.0.0.1:47391/

aif-session

Contributing

Raise an issue with this attachment

aif doctor --bundle > aif-diagnostics.txt

Reporting

You can use these channels

  • Email security@harden.run, or
  • Open a private report via this repository's Security → Report a vulnerability (GitHub Security Advisories).

Include the aif doctor --bundle output.