--- title: "Logfire MCP: review, setup & limitations · Undominated.ai" canonical: https://undominated.ai/mcp-servers/logfire/ description: "Connects assistants to Logfire traces, metrics and observability management tools in the chosen region." --- # Logfire MCP: review, setup & limitations · Undominated.ai > Connects assistants to Logfire traces, metrics and observability management tools in the chosen region. [← Explore all mcp servers](/mcp-servers/) OBSERVABILITY / Pydantic # Logfire MCP Connects assistants to Logfire traces, metrics and observability management tools in the chosen region. [See setup guidance ↓](#setup)[Original source ↗](https://pydantic.dev/docs/logfire/guides/mcp-server/) SOURCE REVIEW Reviewed 2026-10-14 Evidence 2 linked sources Publisher Pydantic Licence Not established [Read what was—and wasn’t—checked ↓](#review) “Pydantic Logfire provides a hosted remote MCP server that you can use without installing anything locally.” [Pydantic · upstream description ↗](https://pydantic.dev/docs/logfire/guides/mcp-server/) Our analysis follows below. 01 / THE REASONING ## Why this made the selection. - Provides region-specific hosted endpoints with browser-based account authorization. - Documents telemetry queries, management tools and organization controls for restricting the available tool surface. - Pydantic’s current guide describes a remote Logfire MCP server with a daily query budget and an option to limit external clients to read-only tools. - The GitHub page listed on this row says the older stdio repository is archived and is no longer being updated. ### A good fit for - Investigating an application trace or telemetry query in Logfire - Querying Logfire through the hosted MCP endpoint while the organization’s query budget remains - Not for installing the archived stdio package as the current server ### Weigh up before choosing - The tool set includes resource writes such as dashboards and alerts. Use organization policy to restrict tools where inspection-only use is required. - Returned telemetry can include attacker-controlled or private content; treat it as data, not instructions. - The earlier pydantic/logfire-mcp repository is archived. Its licence and behavior are not represented as the current hosted implementation. - When the daily budget is used up, Logfire refuses MCP queries until it refills. The page says that refusal does not stop web UI queries or read-token queries. - The archived repository is not the remote server. No git revision was taken from that page. - No licence was stated on the guide, and none was inferred. - This review did not execute the resource. - The server was not called. - MCP query concurrency and daily budgets are separate from ordinary UI access; a working browser session does not establish the agent's query allowance. 02 / THE REVIEW RECORD ## What we actually inspected. Source review has boundaries. A clear record is more useful than a “safe” badge. ### Material inspected - Current hosted-server guide, regional endpoints, authentication, tool permissions and query-budget sections - Archive notice in the previous standalone repository - https://pydantic.dev/docs/logfire/guides/mcp-server/ - https://github.com/pydantic/logfire-mcp - Round 2 reread of acquired publisher setup and permission evidence; see immutable wave review. ### Our findings - The current guide distinguishes hosted regional setup from the older local server and warns that telemetry is untrusted input. - The Logfire guide says there is a concurrency limit and a daily query budget per organization. After the budget is spent, MCP queries are refused until it refills, and that refusal does not stop web UI or read-token queries. - The same guide links a control to limit every external MCP client to read-only tools. - The listed GitHub repository says it is archived, that the stdio server in the package is no longer updated, and that a remote MCP server is the current direction. - The plugin defaults to the US endpoint; organization administrators can restrict external MCP clients to read-only tools and control feedback sharing. ### Not established by this review - No MCP server or upstream package was installed or run. - No authenticated account, client handshake, query, write operation or security test was performed. - The resource was not executed and its tools were not called. - The server was not called. A source review is not a runtime or security certification. The review applies to the material and revision named here. A newer upstream release can change its behavior. 03 / PUT IT TO WORK ## Connect a server deliberately. [Upstream setup instructions ↗](https://pydantic.dev/docs/logfire/guides/mcp-server/) - Use https://logfire-us.pydantic.dev/mcp for US projects or https://logfire-eu.pydantic.dev/mcp for EU projects. - Add the URL as a remote MCP server and complete browser authorization for the intended organization. - Review exposed tools and organizational restrictions before querying sensitive telemetry or making changes. ### Before you start - A Logfire account with access to the target organization and project - An OAuth-capable remote MCP client ### Documented configuration example Use the client and authentication context described in the setup guidance. Configuration formats vary between hosts. { "mcpServers": { "logfire": { "url": "https://logfire-us.pydantic.dev/mcp" } } } Copy configuration ↗ ### Compatibility Remote MCP clients using Streamable HTTP. The guide gives setup for Claude Code, Claude Desktop, Codex, OpenCode, Cursor, VS Code, Cline, Goose, LM Studio and Zed, and for Pi through the community pi-mcp-adapter package. ### Transports Streamable HTTP ### Hosted account connection - Query accessible telemetry and project metadata - Manage supported dashboards, alerts and related resources when allowed by account and organization policy ### Cost model Hosted Logfire service under account, query-budget and resource terms; assistant usage is separate. No hosted implementation licence is established. 04 / FOLLOW THE EVIDENCE ## The source trail. Our notes are separate from the original resource. Check upstream before adopting a new version. - [Official setup and capabilities ↗](https://pydantic.dev/docs/logfire/guides/mcp-server/) Checked 2026-10-14 https://pydantic.dev/docs/logfire/guides/mcp-server/ Supports: summary, upstreamDescription, whySelected, bestFor, install, compatibility, transports, limitations, access, review - [Archived previous standalone implementation ↗](https://github.com/pydantic/logfire-mcp) Checked 2026-10-14 https://github.com/pydantic/logfire-mcp Supports: limitations, review KEEP COMPARING ## Other approaches to consider. Related by category or shared topics. These are alternatives to inspect, not a measured quality order. [### Axiom MCP ↗ Queries Axiom telemetry and manages observability resources through an account-authorized hosted connection.](/mcp-servers/axiom/)[### Checkly MCP Server ↗ Checkly's hosted MCP server over Streamable HTTP: inspect check status and results, test sessions, RCA and result assets, trigger deployed checks, manage monitors and groups, environment variables and status-page incidents. OAuth is recommended; user and service API keys are the documented alternatives.](/mcp-servers/checkly/)[### Datadog MCP ↗ Query Datadog telemetry and operate permitted monitoring resources from an authenticated AI client.](/mcp-servers/datadog/) [AI Tools ↗](/tools/)[Skills ↗](/skills/)[Agents ↗](/agents/)[MCP Servers ↗](/mcp-servers/)[Workflows ↗](/workflows/) ## Continue your investigation - [Inspect resource evidence](/tools/) - [Choose a host](/agents/) - [Inspect reusable workflows](/skills/) - [Use Undominated data](/api/)