Getperspective MCP server
{
"mcpServers": {
"getperspective": {
"type": "http",
"url": "https://getperspective.ai/mcp"
}
}
}{
"mcpServers": {
"getperspective": {
"type": "http",
"url": "https://getperspective.ai/mcp"
}
}
}{
"mcpServers": {
"getperspective": {
"type": "http",
"url": "https://getperspective.ai/mcp"
}
}
}{
"servers": {
"getperspective": {
"type": "http",
"url": "https://getperspective.ai/mcp"
}
}
}{
"mcpServers": {
"getperspective": {
"type": "http",
"url": "https://getperspective.ai/mcp"
}
}
}file .mcp.json (in your project root)file claude_desktop_config.jsonfile ~/.cursor/mcp.jsonfile .vscode/mcp.jsonfile ~/.codeium/windsurf/mcp_config.jsonversion 0.0.8
What it does
This server exposes the Perspective AI platform through 22 tools, allowing MCP clients to create, deploy, analyze, and automate AI conversation agents that replace static forms. It is aimed at developers and product teams who want to manage the full lifecycle of conversational data collection from within tools like Claude Desktop, Cursor, or VS Code.
- Create a lead-qualification conversation agent from a natural-language description and get an embed snippet to place on a pricing page.
- Retrieve and analyze drop-off conversations from the past week to identify why participants abandoned a specific concierge.
- Set up an automation that pushes high-trust-score conversations to HubSpot and sends a Slack notification to a sales channel.
Tools this server exposes
| Tool | Does |
|---|---|
| workspace_list | Lists all workspaces the authenticated user can access. |
| workspace_get | Returns details for a specific workspace by ID. |
| workspace_get_default | Returns the user's default workspace. |
| perspective_list | Lists perspectives in a workspace, or searches by name across all workspaces. |
| perspective_get | Returns full configuration and stats for a specific perspective. |
| perspective_create | Creates a new perspective (conversation agent) from a natural-language brief. |
| perspective_respond | Answers a follow-up question during an ongoing perspective design session. |
| perspective_update | Refines an existing perspective using natural-language feedback. |
| perspective_await_job | Long-polls a perspective design job until it completes. |
| perspective_get_preview_link | Returns a shareable preview URL for testing a perspective before deployment. |
| perspective_get_embed_options | Returns embed code snippets for various embed types and the SDK reference. |
| participant_invite | Creates 48-hour magic-link invites for specific participants, optionally sent via email. |
| perspective_get_stats | Returns aggregate stats and distributions for a perspective over a given time period. |
| perspective_list_conversations | Lists conversations with optional filters for status, trust score, and date range. |
| perspective_get_conversation | Returns full details for a single conversation, including transcript and trust assessment. |
| perspective_get_conversations | Batch-fetches multiple conversations in a token-efficient format for bulk analysis. |
| automation_list | Lists automations on a perspective with their status, channel, and metadata. |
| automation_create | Creates an automation targeting a webhook, email, Slack, or HubSpot channel. |
| automation_update | Updates fields on an existing automation. |
| automation_delete | Permanently deletes an automation. |
| automation_test | Runs an end-to-end test of an automation against a mock conversation. |
| integration_manage | Lists available providers such as Slack and HubSpot, connects them, and searches their tools. |
Before you enable it
An MCP server runs with whatever access you give it. Start with a development account rather than production, and check which tools the server exposes before you let an agent call them unattended.