AI agents (MCP)
MicroSurveys has a remote MCP server, so you can work with your surveys from the AI agent you already use. Ask “what are users saying about checkout?”, pull NPS for last week’s release, or have the agent draft a follow-up survey on the right event.
Server URL: https://console.microsurveys.ai/api/mcp
There’s nothing to install and no key to copy. The first time the agent connects, it opens your browser to sign in to MicroSurveys and asks you to approve it.
Connect
Claude Code
claude mcp add --transport http microsurveys https://console.microsurveys.ai/api/mcpThen run /mcp in Claude Code and choose Authenticate.
The same snippets are in the console under Settings → AI agents, which also lists every connected agent and lets you disconnect one immediately.
What the agent can do
An agent acts as you, in every project you can open in the console.
| Tool | What it does |
|---|---|
list_projects | Your projects (apps) |
list_surveys | Surveys in a project, with status and response counts |
get_survey | A survey’s questions, triggers, targeting and settings |
get_survey_results | Response and completion rates, NPS and averages, distributions, recent comments |
list_responses | Individual responses, paged |
get_survey_insights | AI insights: summary, findings, themes and sentiment, score breakdowns |
list_events / list_user_properties | Your event catalog and user properties, for triggers and targeting |
create_survey | Create a survey. It’s always saved as a draft |
update_survey | Edit a survey’s questions, triggers, targeting or settings |
set_survey_status | Publish, pause, archive or unpublish a survey |
report_issue | Report a bug to the MicroSurveys team, with the survey, platform and SDK version attached |
suggest_feature | Send the MicroSurveys team a feature request |
list_my_requests / get_my_request | Your requests and reports, their status and the team’s replies |
Publishing (set_survey_status → ACTIVE) shows the survey to real users of your app. Agents
are told to confirm with you first, and most clients ask before running it. Editing a live
survey changes it for users straight away too, so clients ask before update_survey as well.
When the agent creates or edits a survey, it gets warnings for any event, property or user property that isn’t in your event catalog, so a typo doesn’t quietly produce a survey that never shows.
Reports and requests sent by an agent go to the same place as Help & feedback in the console, and show up under My requests there. A person on the team reads each one, so agents are told to check the wording with you before sending.
Privacy
- Emails, phone numbers and card numbers in respondents’ answers are replaced with placeholders
(
[email]) before the agent sees them, as with AI insights. End-user properties aren’t shared. - Agents only get access you approve, and only what your own account can reach.
- Access tokens expire after an hour. Disconnecting an agent in Settings stops it at once.