workspace_summary
Summarize service and telemetry activity in the authorized workspace.
OpenTelemetry · Model Context Protocol
Give your coding agent a view of the running application.
Telemio MCP connects compatible AI coding clients to workspace-scoped observability context: traces, logs, metrics, runtimes, dashboards, and integration setup guidance. An agent can investigate live telemetry instead of relying on source code alone.
Hosted streamable HTTP MCP endpoint
https://mcp.telemio.sh/mcp
Authentication is required. Use supported client OAuth authorization or a workspace-scoped MCP key in a trusted secret store. Do not put access tokens in a URL or repository.
Tool capabilities
What the MCP server exposes.
These tool names are listed in the published MCP server card. Actual calls require authorization and connected workspace data.
query_telemetry
Run Telemio TQL queries against workspace telemetry.
list_runtimes
List service, client, worker, and agent runtime sessions.
get_runtime
Inspect a runtime, including its summary and relationships.
list_dashboards
List dashboards accessible in the workspace.
create_dashboard
Create or update TQL-backed visualizations; require explicit approval for writes.
list_integrations
Explore supported integration setup paths.
integration_setup
Retrieve setup guidance for a particular integration.
console_link
Create a console deep link to a telemetry query, runtime, dashboard, or integration.
Reference: machine-readable MCP server card · agent-ready MCP policy. Published capabilities are not a guarantee that a specific client or deployment has connected successfully.
Getting started
Connect without sharing secrets.
Preferred: client OAuth
- Add
https://mcp.telemio.sh/mcpto a compatible MCP client. - Complete the client-initiated OAuth flow using Telemio sign-in and consent.
- Ask the agent for a workspace summary, and confirm that a live MCP tool call succeeds.
OAuth support depends on the MCP client. Use the OAuth/PKCE guide for details.
Alternative: workspace MCP key
- Open Console → MCP and create a workspace key.
- Store the token in your client's trusted secret/environment configuration.
- Connect to the same hosted MCP endpoint and verify a read.
Never embed privileged MCP keys in distributed apps or commit them to source control. A key is scoped to one workspace.
Why this approach
One investigation can cross signals.
OpenTelemetry instruments the application; MCP provides an authorized way for an AI client to query the context that instrumentation produced.
Inspect real behavior
Investigate queries, runtime activity, and available telemetry with actual tool calls. If telemetry is missing, the agent should say so rather than inventing a diagnosis.
Preserve boundaries
Prefer read-only diagnosis, request approval before dashboard changes, avoid transmitting prompt bodies by default, and stay inside the selected workspace.
Frequently asked
Before you connect.
Is this a telemetry collector or an MCP server?
Telemio ingests application signals through OpenTelemetry-compatible paths. The MCP endpoint is a separate, authenticated interface that lets an AI agent investigate the workspace.
Does every MCP client support the OAuth flow?
No. OAuth integration depends on the client. A manual workspace MCP key is available for compatible clients that need an alternative.
Can a coding agent modify data?
The published server card includes create_dashboard. Approve write actions deliberately, and use workspace-scoped credentials with the access needed for the task.
Is there an independent performance benchmark?
Not yet. The public benchmark methodology describes the proposed fair comparison and explicitly distinguishes planned work from measured results.