[VibeDefend] Custom Agent Frameworks Support

Extends VibeDefend beyond coding agents to the agents your teams build themselves. VibeDefend plugs into Claude Agent SDK, OpenAI Agents SDK, Google ADK and LangChain v1 / LangGraph through each framework’s native extension points (hooks, callbacks, middleware). Your security policies are enforced on every agent run: Action Guard inspects and blocks dangerous tool calls, MCP connections go through the MCP Security Proxy, tool outputs are checked for prompt injection, and every agent, tool and MCP server appears in the Agent Inventory, with the same policies and dashboards as your coding agents.