Ship Safe Agents

Scan. Ship. Comply. Verify your AI agents have governance controls before deployment.

EU AI Act Article 14 Deadline: August 2, 2026

Non-compliance penalties: Up to 15M or 3% of global turnover.

What is Agent Governance?

Agent governance ensures AI agents operate within defined boundaries:

Human Oversight - Humans can intervene in high-risk decisions
Authorization Controls - Actions are verified before execution
Audit Trails - All actions are logged for accountability
Least Privilege - Agents have minimal necessary permissions

Inkog verifies these controls are wired correctly before deployment.

The Governance Gap

Traditional security tools scan for code vulnerabilities. But AI agents have a unique risk:


          Traditional Security              Inkog Governance
          ───────────────────               ────────────────
          Code quality                      Human oversight gates
          Authentication                    Authorization wiring
          Data encryption        +          Audit logging config
          Network security                  Permission scoping

Even “secure” code can lack proper governance controls.

What Inkog Detects

Detection	Description	Compliance
Missing Human Oversight	High-risk actions without approval gates	EU AI Act Article 14
Authorization Gaps	Tools executed without permission checks	OWASP LLM06
Missing Audit Logging	Actions not logged for accountability	EU AI Act Article 12
Excessive Permissions	Overly broad tool/data access	EU AI Act Article 15

Supported Platforms

Inkog verifies governance controls across both pro-code and no-code platforms:

Pro-Code Frameworks

Framework	Human Oversight	Authorization	Audit Logging
LangGraph	`interrupt_before`	Custom middleware	Callbacks
CrewAI	`human_input=True`	Tool wrappers	Verbose mode
LangChain	Callbacks	Tool permissions	LangSmith
AutoGen	Approval hooks	Role-based	Logging handlers

No-Code Platforms

Platform	Human Oversight	Authorization	Audit Logging
Copilot Studio	Approval nodes	Azure AD	Power Automate logs
Agentforce	Flow approvals	Permission sets	Event monitoring
n8n	Manual triggers	Credentials	Execution logs
Flowise	-	API keys	-

Quick Start


# Governance-focused scan
inkog . --policy governance
 
# EU AI Act compliance scan
inkog . --policy eu-ai-act

Example: Before and After

Before (Vulnerable)


# Financial trading agent - VULNERABLE
graph.add_edge("analyze", "execute_trade")  # Direct path!
app = graph.compile()

Inkog Finding:


FAIL: universal_missing_oversight
Message: High-risk action 'execute_trade' has no human approval gate
Compliance: EU AI Act Article 14.1, 14.4

After (Compliant)


# Financial trading agent - COMPLIANT
graph.add_edge("analyze", "human_review")
graph.add_edge("human_review", "execute_trade")
app = graph.compile(interrupt_before=["execute_trade"])

Inkog Result:


PASS: All high-risk actions have human oversight gates
Governance Score: 100/100
EU AI Act Readiness: READY

Compliance Evidence

Inkog generates audit-ready compliance reports:


{
  "governance_score": 100,
  "eu_ai_act_readiness": "READY",
  "article_mapping": {
    "Article 14": { "status": "PASS", "description": "Human Oversight" },
    "Article 12": { "status": "PASS", "description": "Record-Keeping" },
    "Article 15": { "status": "PASS", "description": "Robustness" }
  }
}

Use these reports for:

Internal compliance audits
Regulatory submissions
Customer due diligence
Insurance applications

AGENTS.md Governance Scanner

Declare your agent’s capabilities and limitations in AGENTS.md. Inkog cross-validates your declarations against actual code behavior—flagging governance mismatches before deployment.

AGENTS.md


# Customer Support Agent
 
## Capabilities
- Read customer records
- Search knowledge base
- Create support tickets
 
## Limitations
- Read-only database access (no writes)
- No external API calls
- Cannot delete records
 
## Security
- All actions logged
- Human approval for refunds over $100

Learn more about AGENTS.md scanning

Next Steps

AGENTS.md Governance - Declare and validate governance policies
INKOG.md Enhanced Format - Structured governance manifests with compliance scoring
Human Oversight (Article 14) - Configure approval gates
Authorization Verification - Add permission checks
Audit Logging - Enable action logging
Compliance Reports - Generate evidence