armoriq/armorclaw

Registry 信息

项目简介

ArmorIQ 意图执行插件，适用于 OpenClaw AI 代理，旨在与 Claude Skill 集成，确保 AI 代理行为的安全与合规。

要点

• AI 代理的意图执行
• 与 OpenClaw 无缝集成
• 兼容 Claude Skill
• 使用 TypeScript 编写，确保可靠性

使用场景

• 执行 AI 代理行为策略
• 安全的 AI 代理部署
• AI 行为合规监控

README 摘要

# ArmorClaw OpenClaw Plugin Intent-based security enforcement for OpenClaw AI agents. Protect your AI assistant from prompt injection, data exfiltration, and unauthorized tool execution. ## Features - **Intent Verification** - Every tool execution must be part of an approved plan - **Prompt Injection Protection** - Blocks malicious instructions embedded in files - **Data Exfiltration Prevention** - Prevents unauthorized file uploads and data leaks - **Policy Enforcement** - Fine-grained control over tool usage and data access - **Cryptographic Verification** - Optional CSRG Merkle tree proofs for tamper-proof intent tracking - **Fail-Closed Architecture** - Blocks execution when intent cannot be verified ## Installation The recommended path is the one-line installer, which clones OpenClaw, installs the plugin, and writes a working config: ```bash curl -fsSL https://armoriq.ai/install-armorclaw.sh | bash ``` ### Prerequisites - Node.js v22+, pnpm, Git - ArmorClaw API key from [claw.armoriq.ai](https://claw.armoriq.ai) - An LLM provider key (OpenAI, Anthropic, Gemini, or OpenRouter) ### Install (OpenClaw 2026.3.x — no patching required) ```bash openclaw plugins install @armoriq/armorclaw ``` ### Install (OpenClaw 2026.2.x — requires patching) For older OpenClaw versions that need the ArmorClaw runtime patches: ```bash npm install @armoriq/armorclaw@openclaw-2026.2 ``` See the [Quick Start Guide](https://docs.armoriq.ai/docs/installation/quickstart) for details on applying patches for 2026.2.x. ### Verify ```bash openclaw plugins list # Should show: armorclaw | loaded ``` ## Configuration The installer writes this automatically. To review or edit, update `~/.openclaw/openclaw.json`. Endpoints depend on which key flavor you use: - **`ak_claw_…`** keys → A

话题

暂无话题