docs: Add Inkog AI agent security scanning integration

[Benjamin Bandali (cloakmaster)]

Summary

• Add integration page for Inkog, the first security scanner purpose-built for AI agent code
• Inkog scans LangChain & LangGraph projects for behavioral vulnerabilities: prompt injection, infinite loops, token bombing, SQL injection via LLM, missing human oversight
• Covers CLI usage (npx -y @inkog-io/cli scan .), MCP server setup for Claude Desktop/Cursor, GitHub Actions CI integration, and SARIF output for GitHub Security tab
• All findings mapped to OWASP LLM Top 10, EU AI Act, and NIST AI RMF

Why this matters for LangChain developers

Inkog has native framework adapters for LangChain and LangGraph that understand chains, agents, tools, graph nodes, conditional edges, and state -- not just generic Python AST patterns. It helps developers catch security issues during development rather than in production.

Details

• New file: src/oss/python/integrations/tools/inkog.mdx
• Open source: github.com/inkog-io/inkog (Apache 2.0)
• npm packages: @inkog-io/cli, @inkog-io/mcp

Test plan

• Verify the page renders correctly in the docs preview
• Confirm all external links are valid
• Check formatting matches existing integration pages

🤖 Generated with Claude Code

[Benjamin Bandali (cloakmaster)]

Closing this — looking at the integration template and existing pages, this page doesn't fit. Every tool integration under src/oss/python/integrations/tools/ documents a langchain-* Python package with pip install, import, instantiation, and invocation examples. Inkog is an external CLI tool, not a LangChain Python integration.

I'll open an issue to discuss whether security scanning documentation would be useful and where it would best fit in the docs.