Langfuse

Install

Works with Claude Code, Codex, VS Code, Cursor, Windsurf, Claude Desktop. Read the repository before you give any tool access to your data.

npx agentmarket install connector/pulsemcp-com-langfuse

About

No reviews yet. Installed it? Reviews come from verified installers only. How reviews work

Permissions

What this server can do and what it asks for, from its tool list and install config. Checked Oct 25, 2026.

Tools

4 read · 0 write · 0 deleteRead-only

Secrets

2 (API keys, tokens or passwords)

Install scripts

Not scanned yet

Runs as

Node.js package

Tools (4)

Read from the source code; the running server may differ. Descriptions are the server’s own words; hidden characters are shown as [U+…].

• get_observation

  Get full details for a single Langfuse observation by ID. Returns the complete observation including input, output, metadata, model parameters, usage details, cost details, and timing information. Use get_observations or get_trace_detail first to find observation IDs. Any field value exceeding 1000 characters is automatically truncated and saved to a /tmp file. The truncated text includes the file path — use grep on that file to search within large values (e.g. `grep "keyword" /tmp/langfuse_input_abc123.txt`). **Fields included:** id, traceId, type, name, model, modelParameters, input, outp…

• get_observations

  List observations from Langfuse with optional filters and pagination. Returns a summary of each observation: id, traceId, type, name, model, timing, level, and usage/cost metrics. The input and output fields are omitted from the list view for brevity — use get_observation with a specific observation ID to see the full I/O. Typically used with traceId filter to list all observations within a specific trace. Can also be used to search across all traces by type, level, etc. Any field value exceeding 1000 characters is automatically truncated and saved to a /tmp file. The truncated text include…

• get_trace_detail

  Get full details for a single Langfuse trace by ID. Returns the complete trace including input, output, metadata, and all nested observations and scores. This is the detailed view — use get_traces to browse and find trace IDs first. Any field value exceeding 1000 characters is automatically truncated and saved to a /tmp file. The truncated text includes the file path — use grep on that file to search within large values (e.g. `grep "keyword" /tmp/langfuse_input_abc123.txt`). **Observations** are returned as a flat list with parentObservationId for tree reconstruction. Each observation inclu…

• get_traces

  List traces from Langfuse with optional filters and pagination. Returns a summary of each trace: id, name, timestamp, userId, latency, totalCost, tags, and observation/score counts. The input and output fields are omitted from the list view for brevity — use get_trace_detail to see the full trace including I/O. Any field value exceeding 1000 characters is automatically truncated and saved to a /tmp file. The truncated text includes the file path — use grep on that file to search within large values. **Pagination:** Use page and limit to paginate. Response includes meta.totalItems and meta.t…

Signals

Stars

81

Forks

10

Open issues

26

Contributors

4

Last commit (this folder)

Jun 14, 2026 (4 months ago)

Last release

Jun 17, 2026 (4 months ago)

From GitHub (fetched Oct 24, 2026).

Details

Registry name

com.pulsemcp/langfuse

Version

0.1.3

Auth

API key

Packages

npm langfuse-observability-mcp-server

License

MIT

Repository

github.com/pulsemcp/mcp-servers

First seen

Oct 24, 2026

Last checked

Oct 24, 2026

Source

Official MCP Registrygithub-graphql