Inkog
@inkog-io
Inkog MCP Server — the pre-flight check for AI agents
Overview
What is Inkog?
Inkog is a security companion for AI agent development that integrates with MCP-compatible clients such as Claude Desktop, Cursor, and Claude Code. It scans agent code for vulnerabilities, explains findings in plain English, verifies AGENTS.md governance, and audits agent-to-agent delegation—all within the same conversation.
How to use Inkog?
Add Inkog as an MCP server in your client configuration (e.g., claude_desktop_config.json) using npx -y @inkog-io/mcp and set the INKOG_API_KEY environment variable. Then ask your AI assistant to run tools like inkog_scan or inkog_verify_governance with natural language prompts such as “Scan the current directory with Inkog and show me any CRITICAL or HIGH findings.”
Key features of Inkog
- Detects logic flaws and security risks in agent code.
- Verifies AGENTS.md governance matches code behavior.
- Generates compliance reports for EU AI Act, NIST, OWASP.
- Audits MCP servers for tool poisoning and data exfiltration.
- Analyzes multi-agent
