Connect an AI assistant (MCP)
Votinova runs a remote MCP server, so an assistant can read your sessions and results and act on them — through exactly the same API, with exactly the same credential.
What it is
MCP (Model Context Protocol) is the standard AI agents use to call tools. Votinova's server exposes the public API as 23 tools derived from the OpenAPI document: your agent can list presentations, create and open sessions, activate questions or read results — always with your credential and under your limits.
- Create a credential with the permissions the assistant should have. Give it read-only unless it genuinely needs to change things.
- Add the Votinova MCP server to your assistant and paste the credential as its API key.
- Ask it about a session. It discovers what it can do from the same published catalogue this page lists.
Configuration
{ "mcpServers": { "votinova": { "url": "https://mcp.votinova.com/mcp", "headers": { "Authorization": "Bearer vz_live_…" } } }}Works with any MCP client over HTTP transport: Claude, Cursor, VS Code and friends.
Tools are filtered by the credential's REAL scopes: a read-only key does not even see the write tools. Governance and revocation work exactly as for any other API use.
Try it with your agent
- “Create a session on my latest presentation and open it.”
- “How did the second question of yesterday's session turn out?”
- “Subscribe a webhook to question.results.finalized pointing at my backend.”
An assistant is not a separate door. It sees only the capabilities its credential's permissions allow, it counts against the same rate limits, and switching programmatic access off for your organization stops it along with everything else. One capability is deliberately absent: exports are files, and a file is of no use to an assistant — download those through the API or the workspace.
If a key leaks, revoke it. That is the remedy, and it takes effect on the assistant's very next call — nothing is cached and nothing waits for a token to expire.