injection_patterns
Lists built-in injection detection patterns with categories and weights to identify and prevent prompt injection attacks in AI systems.
Instructions
List all built-in injection detection patterns with categories and weights.
Input Schema
TableJSON Schema
| Name | Required | Description | Default |
|---|---|---|---|
No arguments | |||
Output Schema
TableJSON Schema
| Name | Required | Description | Default |
|---|---|---|---|
| result | Yes |
Implementation Reference
- src/agent_safety_mcp/server.py:252-258 (handler)Tool handler for "injection_patterns", which returns the list of injection detection patterns imported from "prompt_shield".
def injection_patterns() -> list[dict]: """List all built-in injection detection patterns with categories and weights.""" from prompt_shield import PATTERNS return [ {"name": p["name"], "category": p["category"], "weight": p["weight"]} for p in PATTERNS ]