AI agents & governance
Building on the MCP server
Connect external AI clients to Planoda over the Model Context Protocol — securely, with the same governance guardrails.
What you'll learn
- Understand what the MCP server exposes and that it shares one tool registry with in-app agents.
- Connect a compatible AI client to call Planoda's tools over a standard interface.
- Confirm the destructive-action broker applies to MCP calls too — governance isn't bypassed.
What the MCP server exposes
6 min — Planoda's tools, available to compatible AI clients.
Planoda ships an MCP (Model Context Protocol) server so external AI clients can call Planoda's tools — read issues, create work, run actions — over a standard interface. The same registry that drives in-app agents drives the MCP surface, so there's one source of truth for what each tool does and whether it's destructive.
Guardrails on the MCP path
5 min — The destructive-action broker applies to MCP calls too.
Calls over MCP run through the same guardrail as in-app agents: the destructive flag comes from the tool registry, sessions lacking destructive capability get a structured approval-required result, and auto-approved operations record an audit row. Governance isn't bypassed by going through MCP.