[Langfuse] Get a specific observation with full details including input, output, usage, and costs.
AI agents call langfuse_get_observation to retrieve information from Shepherd MCP without modifying anything — typically the context-gathering step in research, monitoring, and reporting workflows, before the agent takes action elsewhere.
This tool retrieves observability data about LLM agent sessions from Langfuse. It queries and returns telemetry information (inputs, outputs, token usage, costs) without modifying, deleting, or executing operations. The data returned is historical debugging information, not sensitive credentials or controls.
From the tool's definition Tool description states 'Get a specific observation with full details' - the verb 'Get' and the read-only nature of retrieving observation data (input, output, usage, and costs) without modification or side effects.
Attacks that exploit this kind of access
[Langfuse] Get a specific observation with full details including input, output, usage, and costs. It is categorised as a Read tool in the Shepherd MCP MCP Server, which means it retrieves data without modifying state.
Register the Shepherd MCP server in PolicyLayer and add a rule for langfuse_get_observation: allow, deny, rate-limit, or require approval. Point your MCP client at the PolicyLayer proxy URL and the rule is enforced on every call, before it reaches Shepherd MCP. Nothing to install.
langfuse_get_observation is a Read tool with low risk. Read-only tools are generally safe to allow by default.
Yes. Add a rate_limit block to the langfuse_get_observation rule in your PolicyLayer policy. For example, setting max: 10 and window: 60 limits the tool to 10 calls per minute. Rate limits are tracked per agent session and reset automatically.
Set action: deny in the PolicyLayer policy for langfuse_get_observation. The AI agent will receive a policy violation error and cannot call the tool. You can also include a reason field to explain why the tool is blocked.
langfuse_get_observation is provided by the Shepherd MCP server (neuralis-in/shepherd-mcp). PolicyLayer sits as a proxy in front of this server to enforce policies before tool calls reach the server.
Every MCP server has a record like this.
Type a name, get the same breakdown: verified identity, auth posture, risk grade, capabilities, recommended policy.
Teams ship this data inside their own products. See what a licence covers →