Question 1

What is MCP and what does it let an AI agent do?

Accepted Answer

MCP (Model Context Protocol) is an open standard for connecting AI agents to external tools. With Canviq's MCP server, an agent can deploy a survey, read your PMF score, or pull response data using natural language. No dashboard, no script required.

Question 2

Which AI agents work with the Canviq MCP server?

Accepted Answer

Any MCP-compatible client: Claude (Anthropic), ChatGPT (OpenAI), and custom agents built with the MCP SDK. The Canviq server uses standard HTTP with Server-Sent Events for streaming, so it connects to all major MCP clients out of the box.

Question 3

What can an agent actually do?

Accepted Answer

18 tools across two access tiers. Read tools (available on all plans including Free): get_pmf_score, pmf_segments_list, survey_list, response_list, trigger_list. Write tools (Startup and above): survey_create, survey_publish, survey_pause, survey_archive, survey_update, question_create, question_update, question_delete, questions_reorder, trigger_create, trigger_update, trigger_delete, pmf_create. The read tools enable AI Engine Optimization so AI assistants can surface your PMF data without a paywall.

Question 4

How does authentication work?

Accepted Answer

You create a named agent identity in the Canviq dashboard, issue it an API key with the specific tools it needs, and that's what the agent authenticates with. A dashboard-reading agent gets a read-only key. A survey-creation agent gets a write key. Every action is logged per key.

Question 5

Is the MCP server available on all Canviq plans?

Accepted Answer

MCP is available on Startup and above. The free plan does not include it. Rate limits and how many concurrent agent identities you can create scale with your plan.