Shepherd MCP
Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
| AIOBS_API_KEY | No | Your Shepherd API key | |
| LANGFUSE_HOST | No | Custom Langfuse host URL | https://cloud.langfuse.com |
| AIOBS_ENDPOINT | No | Custom API endpoint URL | |
| LANGFUSE_PUBLIC_KEY | No | Your Langfuse public API key | |
| LANGFUSE_SECRET_KEY | No | Your Langfuse secret API key |
Instructions
Guidance the server publishes about itself, which clients place ahead of the tool catalog so the model reads it before choosing anything.
This server publishes no instructions, or was last inspected before Glama recorded them.
Capabilities
Features and capabilities supported by this server
Protocol revision2025-11-25
| Capability | Details |
|---|---|
| tools | {
"listChanged": false
} |
| experimental | {} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| aiobs_list_sessionsB | [AIOBS] List all AI agent sessions from Shepherd. Returns session metadata, labels, and event counts. |
| aiobs_get_sessionB | [AIOBS] Get detailed information about a specific AI agent session including the full trace tree, LLM calls, function events, and evaluations. |
| aiobs_search_sessionsB | [AIOBS] Search and filter AI agent sessions with multiple criteria including text search, labels, provider, model, function name, date range, errors, and failed evaluations. |
| aiobs_diff_sessionsA | [AIOBS] Compare two AI agent sessions and show their differences including metadata, LLM calls, tokens, latency, providers, models, functions, evaluations, errors, system prompts, request parameters, and response content. |
| langfuse_list_tracesA | [Langfuse] List traces with pagination and filters. Traces represent complete workflows or conversations. |
| langfuse_get_traceB | [Langfuse] Get a specific trace with its observations. Returns full trace data including all observations (generations, spans, events). |
| langfuse_list_sessionsC | [Langfuse] List sessions with pagination. Sessions group related traces together. |
| langfuse_get_sessionB | [Langfuse] Get a specific session with its metrics and aggregated data. |
| langfuse_list_observationsC | [Langfuse] List observations (generations, spans, events) with filters. Observations are the building blocks of traces. |
| langfuse_get_observationB | [Langfuse] Get a specific observation with full details including input, output, usage, and costs. |
| langfuse_list_scoresC | [Langfuse] List scores/evaluations with filters. Scores are attached to traces or observations. |
| langfuse_get_scoreC | [Langfuse] Get a specific score/evaluation with full details. |
| langfuse_search_tracesC | [Langfuse] Search and filter traces with extended criteria including text search, release, cost range, and latency range. Combines API-level and client-side filtering. |
| langfuse_search_sessionsA | [Langfuse] Search and filter sessions with extended criteria including text search, user ID, trace count range, and cost range. Combines API-level and client-side filtering. |
| list_sessionsA | [Deprecated: Use aiobs_list_sessions] List all AI agent sessions from Shepherd. |
| get_sessionA | [Deprecated: Use aiobs_get_session] Get detailed information about a specific AI agent session. |
| search_sessionsC | [Deprecated: Use aiobs_search_sessions] Search and filter AI agent sessions. |
| diff_sessionsB | [Deprecated: Use aiobs_diff_sessions] Compare two AI agent sessions. |
Prompts
Interactive templates invoked by user choice
| Name | Description |
|---|---|
No prompts | |
Resources
Contextual data attached and managed by the client
| Name | Description |
|---|---|
No resources | |
TDQS
Scored across 18 tools
The tools are organized into two clear domains (AIOBS and Langfuse), but within each domain there is significant overlap. For example, aiobs_list_sessions and aiobs_search_sessions both retrieve sessions, with the latter adding filtering, which could cause confusion about which to use. Similarly, langfuse_list_sessions and langfuse_search_sessions serve very similar purposes. The deprecated tools further add redundancy without clear differentiation from their AIOBS counterparts.
The naming follows a consistent prefix_verb_noun pattern (e.g., aiobs_diff_sessions, langfuse_get_observation), which is predictable and readable. However, there are minor deviations: some tools use 'list' while others use 'search' for similar operations, and the deprecated tools lack the prefix, breaking full consistency but not severely impacting usability.
With 18 tools, the count is borderline high for the scope, which appears to be monitoring and analyzing AI agent sessions and traces. The inclusion of deprecated tools (6 out of 18) inflates the number unnecessarily, making the set feel heavy and cluttered, though the core functionality is well-represented.
The tool set provides comprehensive coverage for querying and analyzing sessions, traces, observations, and scores across two systems (AIOBS and Langfuse), with operations like get, list, search, and diff. Minor gaps might include update or delete operations, but these are likely not needed for the monitoring domain, and agents can work effectively with the provided read-oriented tools.