The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill downloads and runs the langfuse-mcp package from PyPI using uvx, which is the standard delivery method for this tool.
[COMMAND_EXECUTION]: It configures the agent to execute shell commands that launch the Langfuse service as a subprocess for tool interaction.
[DATA_EXFILTRATION]: The skill facilitates data transfer to the official Langfuse domain (https://cloud.langfuse.com) to manage observability traces and application logs.
[PROMPT_INJECTION]: This skill has an indirect prompt injection surface. Ingestion points: Untrusted data from the Langfuse API enters the context via fetch_traces and get_prompt in SKILL.md. Boundary markers: Absent. Capability inventory: The skill can execute subprocesses via uvx langfuse-mcp and write files to /tmp as noted in SKILL.md. Sanitization: Absent.
[SAFE]: The instructions correctly advise users to store API keys in environment variables and to exclude sensitive configuration files from version control using .gitignore.

langfuse