Guides
Connect an AI agent to Kebivo
When an agent can't answer a question reliably, it can report it to your self-hosted Kebivo inbox. Each guide runs an example from the repository's examples/ directory and says what was tested.
Framework integrations
Framework integration
OpenAI Agents SDK
Connect a Python OpenAI Agents SDK agent to Kebivo over MCP and verify gap reports in the inbox.
Framework integration
LangGraph with LangChain
Add Kebivo MCP tools to a LangGraph agent via LangChain, then verify gap reports in the inbox.
Before you start
Every guide needs a running Kebivo and a project token with the feedback:gap scope. The quickstart creates both. The connection check in each guide runs without a model or model API key. A full agent run needs a configured model.
The agent decides whether to call report_knowledge_gap. When the call succeeds, Kebivo stores the report durably, then groups and routes it asynchronously. Look for the event in the inbox and for any delivery in Deliveries. An accepted report may not have reached a destination yet, and it doesn't change the agent's source knowledge.
If a question must always be reported, don't leave it to the model: submit it from application code with the REST API. The MCP guide covers token scopes, input limits, errors and transport restrictions.