Snowflake has released a redesigned Observe MCP server and new CLI tool that let AI agents directly query observability data without going through a UI or an LLM intermediary, making incident triage and debugging faster and cheaper.
Engineers can now build custom alert-response agents and copilots that access the same production telemetry APIs and workflows humans use, with prebuilt skills for common observability tasks available immediately.
What happened
Snowflake announced general availability of a redesigned Observe MCP server and a new Observe CLI with full feature parity, enabling AI agents and engineers to query observability data directly through APIs, model context protocol, and command-line interfaces instead of through the UI alone.
Why it matters
As AI agents increasingly read and act on operational telemetry, observability platforms must support both human and machine access. The new tools let coding assistants investigate errors, AI SREs correlate failures across services, and engineers build custom alert-triage agents that automatically pull production context—without the latency and cost of a language-model intermediary.
What to watch
The redesigned MCP server removes the LLM intermediary built into the original version, giving agents direct access to Observe's APIs and context graph; general availability is rolling out now across all clusters, with eu-2 and ca-1 clusters to follow shortly. Both tools ship with prebuilt skills for incident investigation, failure tracing, and change validation.
Snowflake announced general availability of a redesigned Observe MCP server and a new Observe CLI, both of which provide programmatic access to observability data for AI agents and engineers. The announcement reflects a fundamental shift in how observability platforms are used: agents, not just humans, are now primary consumers of operational telemetry. Coding assistants can investigate errors before engineers open tickets, and AI SREs can correlate failures across services without human intervention.
The new tools are built around the principle that observability must work for both humans and machines. The previous Observe CLI covered only a narrow slice of platform functionality, with most features accessible only through the UI. The redesigned CLI now provides agent-compatible programmatic access to the entire Observe platform from the command line, with the same surface available through the MCP server and the UI. It works in agent environments like Claude Code as well as interactive terminal sessions. Some workflows run autonomously in the background for routine tasks; others are designed for interactive use, where an engineer guides an investigation directly. The CLI serves as a programmatic interface for composing, automating, and extending observability workflows.
The most significant architectural change is the removal of the LLM intermediary from the MCP server. In the old design, agents sent queries to a single endpoint where Observe's custom LLM interpreted the question and returned an answer. This added latency and cost for every agent call. In the new architecture, agents connect directly to Observe's APIs—the same APIs that power the frontend—and have direct access to Observe's context graph. This allows agents to determine which data sets are most valuable to query, write efficient OPAL (Observe's query language), and run precise observability queries against any telemetry data without intermediary overhead. Through the MCP server, agents can access APM service maps, OpenTelemetry data collection setup, and active alert listings.
Both the CLI and MCP server ship with a library of prebuilt skills: structured workflows based on how Observe engineers solve common observability tasks. These skills enable agents and engineers to investigate incidents, trace failures, validate changes, find n+1 issues, and detect outliers, with no additional setup required after configuration. General availability is rolling out now across all clusters, with eu-2 and ca-1 clusters to follow shortly. To get started, users log into their Observe tenant, navigate to "Manage account" → "MCP server details", and follow instructions for their agent of choice.
The shift toward AI agents as primary consumers of operational data is reshaping how observability platforms must function. Historically, telemetry systems were built first for human operators reading dashboards; now, coding assistants, SREs, and alert-handling agents need programmatic, low-latency access to the same data. Snowflake's redesign addresses this by stripping out the cost and latency overhead of an embedded language model that once mediated all agent queries. Instead of an agent asking a question in natural language, which a custom LLM would interpret and answer, the agent now calls Observe's APIs directly—the same APIs that power the web UI. This architectural change reflects broader industry movement: as foundation models improve, the need for domain-specific LLM layers diminishes, making direct API access faster and cheaper.
The availability of full parity between the CLI and MCP server means engineers and agents operate on identical tooling, reducing fragmentation and the need for custom workarounds. The article notes that several teams had already built unofficial integrations because programmatic access was missing; the new tools are designed to cover those use cases (dashboard updates, chart exports, APM error queries) in a supported, efficient way. The inclusion of prebuilt skills—structured workflows for incident investigation, failure tracing, and outlier detection—lowers the barrier to entry: both humans and machines can operate these workflows immediately without custom setup.
AI-summarized, only the topics you pick — one digest a day via Email, Slack, or Discord.
Free · takes 30 seconds · unsubscribe anytime
Ask AI anything about this article. Q&As are published on this page for other readers too.
Researchers from Google, the University of Chicago, and other institutions disabled the internal mechanism in…
Observe by Snowflake announced general availability of a redesigned MCP (model context protocol) server and a…
Apple is taking a focused approach to AI by integrating Apple Intelligence into existing products rather than…
Microsoft's Vice President for Southern Europe, Charles Calestroupat, told Fortune Greece that the region—Gree…
Anthropic's biological and chemical weapons classifiers—filters designed to block dangerous knowledge extracti…
A survey by Epoch AI and Ipsos of 1,106 employed US adults (conducted July 10–19, 2026) found that 20 percent…
The AI news that matters, in one minute each morning.
Sign up free