generated: '2026-07-20'
method: searched
status: published
source: https://www.polarsignals.com/docs/mcp
blog: https://www.polarsignals.com/blog/posts/2025/07/17/the-mcp-for-performance-engineering
server:
  name: polar-signals
  transport: http
  url: https://api.polarsignals.com/api/mcp/
  auth: Bearer token in the Authorization header (generate from project settings in Polar Signals Cloud,
    or the "AI" section of the dashboard); connects an MCP client against the user's projects (service-account
    / OAuth session).
  clients:
    claude_code: 'claude mcp add --transport http polarsignals https://api.polarsignals.com/api/mcp/ --header
      "Authorization: Bearer <TOKEN>"'
    cursor: '{"mcpServers":{"polarsignals":{"url":"https://api.polarsignals.com/api/mcp/","headers":{"Authorization":"Bearer
      <TOKEN>"}}}}'
description: Official Model Context Protocol server that lets AI assistants (Claude, Cursor, Copilot,
  Gemini) query Polar Signals continuous-profiling data using natural language, discover projects and
  profile types, explore labels, and pull profiles via PromQL-style selectors.
tools:
- name: get_projects
  description: List the projects the authenticated user can access.
- name: profile_types
  description: Discover the available profile types for a project (cpu, memory, gpu, ...).
- name: labels
  description: Explore the metadata label keys available on profiling data.
- name: values
  description: Get the values for a specific label key.
- name: get_profile
  description: Query and retrieve a profile using PromQL-style label selectors.
deployment:
  mode: remote
  endpoint: https://api.polarsignals.com/api/mcp/
  verified: probed
  probe: gated
  checked: '2026-08-12'
  source: catalog MCP census