Skip to content

PolicyShield

Quick Start

mishabar410/PolicyShield

Quick Start¶

1. Initialize a project¶

policyshield init --preset security --no-interactive

This creates: - policies/rules.yaml — Your rule definitions - tests/test_rules.yaml — Auto-generated test cases - policyshield.yaml — Configuration file

2. Review your rules¶

# policies/rules.yaml
shield_name: security-policy
version: 1

rules:
  - id: block-shell-injection
    description: "Block destructive shell commands"
    when:
      tool: exec
      args_match:
        command:
          regex: "rm\\s+-rf|mkfs|dd\\s+if="
    then: block
    severity: critical
    message: "Destructive shell commands are not allowed."

3. Validate and lint¶

policyshield validate policies/
policyshield lint policies/rules.yaml

4. Run tests¶

policyshield test tests/

5. Use in your agent¶

from policyshield.shield.engine import ShieldEngine

engine = ShieldEngine(rules="policies/rules.yaml")

# Check a tool call
result = engine.check("exec", {"command": "rm -rf /"})
print(result.verdict)  # Verdict.BLOCK
print(result.message)  # "Destructive shell commands are not allowed."

# Safe calls pass through
result = engine.check("web_search", {"query": "best restaurants"})
print(result.verdict)  # Verdict.ALLOW

6. Or start the HTTP server¶

pip install "policyshield[server]"
policyshield server --rules policies/rules.yaml --port 8100

curl -X POST http://localhost:8100/api/v1/check \
  -H "Content-Type: application/json" \
  -d '{"tool": "exec", "args": {"command": "rm -rf /"}}'
# → {"verdict": "BLOCK", "message": "Destructive shell commands are not allowed."}

Next steps¶