# Vaultak

Runtime security for AI agents. Score your agent's risk level across 5 dimensions, check if actions should be allowed or blocked, and get ready-to-use security policies. Zero configuration required.

## Quick Start

```bash
# Connect this server (installs CLI if needed)
npx -y @smithery/cli@latest mcp add samueloladji/Vaultak

# Browse available tools
npx -y @smithery/cli@latest tool list samueloladji/Vaultak

# Get full schema for a tool
npx -y @smithery/cli@latest tool get samueloladji/Vaultak vaultak_risk_score

# Call a tool
npx -y @smithery/cli@latest tool call samueloladji/Vaultak vaultak_risk_score '{}'
```

## Direct MCP Connection

Endpoint: `https://vaultak--samueloladji.run.tools`

## Tools (3)

- `vaultak_risk_score` — Score an AI agent's risk level across 5 security dimensions: action type, resource sensitivity, blast radius, behaviora…
- `vaultak_policy_check` — Check whether a specific agent action should be ALLOWED or BLOCKED based on Vaultak's policy engine. Evaluates action/r…
- `vaultak_get_policy_templates` — Get ready-to-use Vaultak policy templates for common AI agent security scenarios: database protection, file system limi…

```bash
# Get full input/output schema for a tool
npx -y @smithery/cli@latest tool get samueloladji/Vaultak <tool-name>
```