# agent-safety-mcp

> Use this tool when you need to ensure the safety and security of AI interactions, detecting prompt injection and scanning for leaked secrets or personally identifiable information (PII) in large language model (LLM) outputs. It solves problems related to data protection, cost management, and audit trail creation, accepting LLM outputs and API requests as inputs and producing signed audit trails and budget enforcement as outputs. Ideal for use in applications where sensitive information is involved and security is a top priority.

Canonical page: https://skillsregistry.net/skills/luciferforge-agent-safety-mcp  
JSON: https://api.skillsregistry.net/v1/skills/luciferforge-agent-safety-mcp

## Description

Unified MCP safety server that detects prompt injection (75 patterns), scans LLM outputs for leaked secrets/PII, enforces API cost budgets, and creates signed audit trails.  Zero ML dependencies, pure Python.

## Trust

- **Trust score (0–1):** 0.94
- **Verification tier:** verified
- **Last scanned:** 2026-09-28

## Facts

- **Version:** 1.0.0
- **Skill type:** atomic
- **Execution layer:** mcp-remote
- **Runtime environment:** api
- **Category:** ai-ml
- **Updated:** 2026-09-28

## Source

- **Source listing:** [Glama](https://glama.ai/mcp/servers/au9iotjsgu)
- **Repository:** <https://github.com/LuciferForge/agent-safety-mcp>

## Use it

Resolve this record through the SkillsRegistry MCP server (no auth, read-only):

```
claude mcp add --transport http --scope user skillsregistry https://api.skillsregistry.net/mcp
```

```json
{
  "jsonrpc": "2.0",
  "id": 1,
  "method": "tools/call",
  "params": {
    "name": "get_skill",
    "arguments": {
      "slug": "luciferforge-agent-safety-mcp"
    }
  }
}
```

REST: `GET https://api.skillsregistry.net/v1/skills/luciferforge-agent-safety-mcp` · pull for local use: `GET https://api.skillsregistry.net/v1/skills/luciferforge-agent-safety-mcp/pull`

---
SkillsRegistry indexes agent skills from public registries and GitHub. Skills we have analysed are scanned with Circle-IR and scored on six dimensions; each listing states its scan coverage. More: https://skillsregistry.net/llms.txt