/agent-governance

name: agent-governance
description: |
  Patterns and techniques for adding governance, safety, and trust controls to AI agent systems. Use this skill when:
  - Building AI agents that call external tools (APIs, databases, file systems)
  - Implementing policy-based access controls for agent tool usage
  - Adding semantic intent classification to detect dangerous prompts
  - Creating trust scoring systems for multi-agent workflows
  - Building audit trails for agent actions and decisions
  - Enforcing rate limits, content filters, or tool restrictions on agents
  - Working with any agent framework (PydanticAI, CrewAI, OpenAI Agents, LangChain, AutoGen)

Agent Governance Patterns

Patterns for adding safety, trust, and policy enforcement to AI agent systems.

Overview

Governance patterns ensure AI agents operate within defined boundaries — controlling which tools they can call, what content they can process, how much they can do, and maintaining accountability through audit trails.

User Request → Intent Classification → Policy Check → Tool Execution → Audit Log
                     ↓                      ↓               ↓
              Threat Detection         Allow/Deny      Trust Update

When to Use

• **Agents with tool access**: Any agent that calls external tools (APIs, databases, shell commands)
• **Multi-agent systems**: Agents delegating to other agents need trust boundaries
• **Production deployments**: Compliance, audit, and safety requirements
• **Sensitive operations**: Financial transactions, data access, infrastructure management

---

Pattern 1: Governance Policy

Define what an agent is allowed to do as a composable, serializable policy object.

from dataclasses import dataclass, field
from enum import Enum
from typing import Optional
import re

class PolicyAction(Enum):
    ALLOW = "allow"
    DENY = "deny"
    REVIEW = "review"  # flag for human review

@dataclass
class GovernancePolicy:
    """Declarative policy controlling agent behavior."""
    name: str
    allowed_tools: list[str] = field(default_factory=list)       # allowlist
    blocked_tools: list[str] = field(default_factory=list)       # blocklist
    blocked_patterns: list[str] = field(default_factory=list)    # content filters
    max_calls_per_request: int = 100                             # rate limit
    require_human_approval: list[str] = field(default_factory=list)  # tools needing approval

    def check_tool(self, tool_name: str) -> PolicyAction:
        """Check if a tool is allowed by this policy."""
        if tool_name in self.blocked_tools:
            return PolicyAction.DENY
        if tool_name in self.require_human_approval:
            return PolicyAction.REVIEW
        if self.allowed_tools and tool_name not in self.allowed_tools:
            return PolicyAction.DENY
        return PolicyAction.ALLOW

    def check_content(self, content: str) -> Optional[str]:
        """Check content against blocked patterns. Returns matched pattern or None."""
        for pattern in self.blocked_patterns:
            if re.search(pattern, content, re.IGNORECASE):
                return pattern
        return None

Policy Composition

Combine multiple policies (e.g., org-wide + team + agent-specific):

def compose_policies(*policies: GovernancePolicy) -> GovernancePolicy:
    """Merge policies with most-restrictive-wins semantics."""
    combined = GovernancePolicy(name="composed")

    for policy in policies:
        combined.blocked_tools.extend(policy.blocked_tools)
        combined.blocked_patterns.extend(policy.blocked_patterns)
        combined.require_human_approval.extend(policy.require_human_approval)
        combined.max_calls_per_request = min(
            combined.max_calls_per_request,
            policy.max_calls_per_request
        )
        if policy.allowed_tools:
            if combined.allowed_tools:
                combined.allowed_tools = [
                    t for t in combined.allowed_tools if t in policy.allowed_tools
                ]
            else:
                combined.allowed_tools = list(policy.allowed_tools)

    return combined


# Usage: layer policies from broad to specific
org_policy = GovernancePolicy(
    name="org-wide",
    blocked_tools=["shell_exec", "delete_database"],
    blocked_patterns=[r"(?i)(api[_-]?key|secret|password)\s*[:=]"],
    max_calls_per_request=50
)
team_policy = GovernancePolicy(
    name="data-team",
    allowed_tools=["query_db", "read_file", "write_report"],
    require_human_approval=["write_report"]
)
agent_policy = compose_policies(org_policy, team_policy)

Policy as YAML

Store policies as configuration, not code:

# governance-policy.yaml
name: production-agent
allowed_tools:
  - search_documents
  - query_database
  - send_email
blocked_tools:
  - shell_exec
  - delete_record
blocked_patterns:
  - "(?i)(api[_-]?key|secret|password)\\s*[:=]"
  - "(?i)(drop|truncate|delete from)\\s+\\w+"
max_calls_per_request: 25
require_human_approval:
  - send_email

import yaml

def load_policy(path: str) -> GovernancePolicy:
    with open(path) as f:
        data = yaml.safe_load(f)
    return GovernancePolicy(**data)

---

Pattern 2: Semantic Intent Classification

Detect dangerous intent in prompts before they reach the agent, using pattern-based signals.

from dataclasses import dataclass

@dataclass
class IntentSignal:
    category: str       # e.g., "data_exfiltration", "privilege_escalation"
    confidence: float   # 0.0 to 1.0
    evidence: str       # what triggered the detection

# Weighted signal patterns for threat detection
THREAT_SIGNALS = [
    # Data exfiltration
    (r"(?i)send\s+(all|every|entire)\s+\w+\s+to\s+", "data_exfiltration", 0.8),
    (r"(?i)export\s+.*\s+to\s+(external|outside|third.?party)", "data_exfiltration", 0.9),
    (r"(?i)curl\s+.*\s+-d\s+", "data_exfiltration", 0.7),

    # Privilege escalation
    (r"(?i)(sudo