A security guardrail package for AI coding agents checks agent activity for prompt-injection attacks, unsafe web content, data leaks, and destructive commands. The listed versions support Claude Code, including a self-hosted option.

Use it to scan web content, block data exfiltration, detect jailbreaks and prompt injection, and require authorization for destructive commands in Claude Code.

It helps prevent untrusted instructions or agent actions from causing security problems, exposing data, or damaging files and systems. The self-hosted version keeps enforcement in local hooks and avoids accidental contact with the hosted service.

agentguards — Plugin by alelaguard

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

Claude Code

/plugin marketplace add alelaguard/agentguards-plugins

agentmods

npx agentmods add plugins/alelaguard/agentguards-plugins/marketplace

Clone the repo

git clone --depth 1 https://github.com/alelaguard/agentguards-plugins

Made for: Claude Code.

Per session not measured What this adds to a session before it is invoked.

When invoked not measured Not applicable: nothing here is loaded into a session.

Security scan A 0 findings. Scan, not verified.

Origin original No closer match found in the catalogue.

Security

Grade A, and why

agentguards scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured yesterday.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

.claude-plugin/marketplace.json · 39 lines

What it actually says

{
  "name": "agentguards",
  "description": "Official AgentGuards marketplace — LLM security guardrails for AI coding agents.",
  "owner": {
    "name": "AgentGuards",
    "url": "https://agentguards.co"
  },
  "plugins": [
    {
      "name": "agentguards-claude",
      "source": "./claude",
      "description": "LLM security guardrails for Claude Code: jailbreak/prompt-injection detection, web-content scanning, data-exfiltration blocking, and destructive-command authorization.",
      "category": "Developer Tools",
      "keywords": [
        "security",
        "guardrails",
        "prompt-injection",
        "jailbreak-detection",
        "llm-security",
        "mcp"
      ]
    },
    {
      "name": "agentguards-claude-selfhosted",
      "source": "./claude-selfhosted",
      "description": "LLM security guardrails for Claude Code, for a self-hosted AgentGuards appliance: jailbreak/prompt-injection detection, web-content scanning, data-exfiltration blocking, and destructive-command authorization, enforced entirely by hooks (no bundled MCP server, so it never talks to AgentGuards' hosted service by accident).",
      "category": "Developer Tools",
      "keywords": [
        "security",
        "guardrails",
        "prompt-injection",
        "jailbreak-detection",
        "llm-security",
        "self-hosted"
      ]
    }
  ]
}

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

yesterday First seen · 39 lines scan A 88b2ab9aa86b

Subscribe to this mod's changes

agentguards is a plugin published in the GitHub repository alelaguard/agentguards-plugins (1 stars, last pushed 2d ago), licensed MIT. Its token cost is not measured: this kind of file is read by the harness, not the model. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.

Data and AI Guardrails Security Writing and docs ai-security claude-code claude-code-plugin codex llm-security

agentguards

Getting it into your agent

Grade A, and why

Nothing flagged

What it actually says

What this file has done since we first saw it

Other plugins, from other repositories

unboundcompute

trust-issues

trust-issues

cordon

bridgeward

gauntlex