LangSmith (MCP Server)
Last updated September 23, 2025
About LangSmith (MCP Server)
Unified observability and evaluation platform for AI apps with an MCP server for history and prompts.
LangSmith is a product by LangChain that provides teams with tools to debug, evaluate, and monitor the performance of AI applications — including agents built with or without the LangChain framework. It offers tracing of application behavior, evaluation with human feedback or automated “LLM-as-judge”, prompt testing/versioning, dashboarding, and alerting. It supports features like self-hosting for enterprises, controlling where data is stored, and monitoring business-relevant metrics.
Key features
- Tracing and detailed observability of LLM/agent runs
- Automated and human-in-the-loop evaluations
- Prompt engineering with versioning and testing
- Dashboards and monitoring for latency, cost, and quality
- Self-hosting and hybrid deployment options
- OpenTelemetry-compliant, API-first platform
- MCP server for conversation history, prompts, search, and analytics
Related tools
An MCP server that connects AI assistants with SigNoz observability data for metrics, alerts, dashboards, and service monitoring.
An MCP server that connects AI assistants with Axiom datasets via Axiom Processing Language (APL).
An MCP server that lets AI assistants query AgentOps traces and spans for debugging and observability.
MCP server for Sentry integration, enabling AI assistants to access error tracking, issue management, and debugging workflows.
MCP server for Grafana integration, enabling AI assistants to access dashboards, metrics, and observability data.
