HeronSignal MCP lets tools like Cursor use monitoring context while helping developers investigate and fix performance or frontend issues.
HeronSignal MCP gives agents production evidence without exposing raw telemetry. Tokens are read-only by default; write scopes are narrow, explicit opt-ins.
MCP tokens are scoped to one workspace so coding agents only see the monitoring context they need.
Agents can read AI analysis, page performance, frontend errors and logs, session journeys, traffic summaries, funnel results with conversion trends, health status, alarms, and per-issue trends to verify a fix worked.
Optional per-token scopes let an agent create funnels or metrics and triage issues (resolve, ignore, reopen), shared with your team's Logger. No scope updates your repository or production code; fixes stay in the workflow you control.
HeronSignal returns summaries and paginated details instead of overwhelming agents with huge telemetry payloads.
Agents can connect code changes to real user pain, such as slow pages, broken clicks, failed APIs, or JavaScript errors.
One install on your site, one token in your editor. No infrastructure to maintain.
Add one script tag to your HTML. HeronSignal begins capturing visitor telemetry, errors, and performance data.
From your workspace settings, create an MCP token scoped to one workspace: read-only by default, with optional write scopes you choose.
Paste the server block into your Cursor, Claude Desktop, or compatible MCP client config file.
Ask your agent to investigate a slow page, a spike in errors, or a post-deploy regression. It reads the live data automatically.
Install HeronSignal, generate an MCP token, and your agent gains live access to performance, errors, sessions, funnels, health, and alarms.