What is the main technical change in the new Observe MCP server?

The original MCP server routed agent queries to a custom LLM that interpreted questions and returned answers. The new architecture removes the LLM middleman, allowing agents to connect directly to Observe's APIs and query the context graph themselves, reducing latency and operational cost.

What can agents and engineers do with the new CLI and MCP server?

Both tools provide the same programmatic access to Observe's capabilities, including APM service maps, OpenTelemetry data collection, and active alert listing. Engineers can build custom alert-triage agents that automatically query Observe when alerts fire, create incident-investigation copilots, update dashboards after investigations, fetch chart images for reports, and query APM errors directly.

When will the new MCP server and CLI be available?

General availability is rolling out now across all clusters, with eu-2 and ca-1 regions to follow shortly.

Large Language ModelsAI Coding AssistantsSnowflake AI BlogAug 13, 2026Aug 14, 2026, 06:01 JST

Snowflake's Observe Launches MCP Server and CLI for AI Agents

Send on LINE

3 Key Points

What happened
Snowflake announced general availability of a redesigned Observe MCP (model context protocol) server and a new Observe CLI with full parity between the two tools. The MCP server now connects AI agents directly to Observe's APIs instead of routing queries through an LLM intermediary, and both tools expose the same operations for agents and engineers.
Why it matters
Teams are building custom alert-triage agents and incident-investigation copilots that query Observe automatically when alerts fire or during investigations. By removing the LLM middleman and giving agents direct API access, Observe reduces latency and operational cost while letting engineers and AI agents access the same observability capabilities—APM service maps, OpenTelemetry data, and active alerts—through CLI, MCP, or the UI.
What to watch
General availability is rolling out now across all clusters, with eu-2 and ca-1 regions to follow shortly. Both the CLI and MCP server include a library of prebuilt skills for common observability workflows (incident investigation, failure tracing, change validation, n+1 detection, and outlier detection) that work immediately after configuration.

Not sure about something? Ask the AI

Questions and answers are published on this page.

Summaries like this, in your inbox every morning.

Get news like this every morning, free

Context & Analysis

The shift reflects a broader trend in AI observability: as agents become the primary consumers of telemetry data—not humans at a dashboard—observability platforms must expose their data through APIs and automation-friendly protocols, not just user interfaces. Snowflake's decision to remove the LLM intermediary is a pragmatic response to advances in foundation models; modern LLMs no longer need Observe to interpret queries on their behalf, and routing through an intermediary layer added cost and latency that customers and engineering teams had already worked around with custom workarounds.

The prebuilt skills and API-first design lower the barrier to entry for engineers building AI-native incident response: teams can now compose and automate observability workflows without building their own integrations or losing the fine-grained data visibility that direct API access provides. By ensuring feature parity between CLI, MCP, and UI—so every operation available to an agent is also available to a human—Snowflake positions Observe for hybrid human-and-agent workflows rather than replacing human judgment with automation.

FAQ

What is the main technical change in the new Observe MCP server?: The original MCP server routed agent queries to a custom LLM that interpreted questions and returned answers. The new architecture removes the LLM middleman, allowing agents to connect directly to Observe's APIs and query the context graph themselves, reducing latency and operational cost.
What can agents and engineers do with the new CLI and MCP server?: Both tools provide the same programmatic access to Observe's capabilities, including APM service maps, OpenTelemetry data collection, and active alert listing. Engineers can build custom alert-triage agents that automatically query Observe when alerts fire, create incident-investigation copilots, update dashboards after investigations, fetch chart images for reports, and query APM errors directly.
When will the new MCP server and CLI be available?: General availability is rolling out now across all clusters, with eu-2 and ca-1 regions to follow shortly.

Snowflake AI BlogRead Original Article

Send on LINE

AI news that matters for your work, delivered every morning.

Pick your industry and the AI tools you use, and get news related to your work every day.

Get it free

Free · 30 seconds with Google · unsubscribe anytimeWhat is AIToday? →

Ask AI

Ask AI anything about this article. The AI reads this article, earlier AIToday articles, and Wikipedia, and cites its sources. Q&As are published on this page for other readers too.