Groundcover opens Agent Mode to Claude

Groundcover said on Oct. 14, 2026 that external coding agents including Anthropic’s Claude and OpenAI’s Codex can hand production investigations to its observability agent, a move that lets AI coding tools trigger deeper incident analysis inside Groundcover’s own system.

The launch builds on Agent Mode, Groundcover’s observability workflow that the company says can investigate logs, metrics, traces and events through a single gcQL query language, then return either a completed root cause analysis or a draft dashboard depending on what the user asked for. Groundcover says the system uses cross-signal identifiers attached at ingest to connect telemetry across signals, and that it can run as background jobs to generate topology maps, daily incident summaries and suggested configuration changes for human review.

What does Groundcover's Agent Mode do?

Groundcover says Agent Mode is an observability agent that can take open-ended investigations, including cases with no pre-existing monitor, incident ticket or known failure state. The company says engineers can access it from any page in the platform, open multiple tabs for parallel investigations and use it through connectors in collaboration tools such as Slack and Linear.

The system uses gcQL as a unified query language across logs, metrics, traces and events, and Groundcover says it pushes complex queries to the backend instead of pulling raw data into the model’s context window. Groundcover also says Agent Mode can create first-class platform assets including dashboards, monitors, gcQL queries and OTTL pipelines.

How does Groundcover connect AI coding agents and incident response?

Groundcover’s connector documentation describes a workflow in which an engineer diagnoses a latency spike or other issue in Groundcover and then asks a coding agent such as Cursor, GitHub Copilot, Claude or Codex to push the fix from the same conversation. SiliconANGLE reported that the new launch also works inside Slack incident threads, where teams can mention @groundcover and let the agent use the conversation as context plus connected systems such as Linear and GitHub.

Groundcover said it first connected Agent Mode to Slack, Linear and GitHub in June 2026. The company’s connector docs also say admins control which connectors the organization can use, while engineers connect their own accounts, and that Agent Mode acts in connected tools using the engineer’s credentials and within that engineer’s permissions.

Groundcover says role-based access control scopes connector authorization to admin-established levels, and disabling a connector stops Agent Mode from using it for everyone without deleting existing credentials. The company’s examples show a one-conversation flow from incident diagnosis to code change, reflecting a tighter loop between observability and delivery.

How does Groundcover say the system handles data and access?

Groundcover says Agent Mode can run on Amazon Bedrock, OpenAI ChatGPT or Anthropic Claude inside the customer’s own cloud account. The company says prompts, logs and traces do not leave that infrastructure, and that token costs are paid directly by the customer at cost with no Groundcover markup.

Groundcover also says customers can set quota budgets per user or team for Agent Mode usage. Separately, SiliconANGLE reported that Groundcover uses a single sensor to capture application and infrastructure behavior without requiring every service to be instrumented, and that the company charges customers by the node rather than by data volume, so retaining more telemetry at higher fidelity does not increase the bill.

Groundcover’s product documentation says it deploys an eBPF sensor at the kernel for deep automatic telemetry without requiring developer instrumentation. The company says every collected signal is enriched with a cross-signal identifier at ingest.

What else is Groundcover shipping around AI observability?

Groundcover also documents telemetry integrations for AI coding tools and agent frameworks including Claude Code, Claude Cowork, OpenAI Codex, OpenClaw and LangChain. Its observability docs say those integrations cover signals such as cost, usage, errors and latency, which suggests the company is positioning Agent Mode inside a broader stack for monitoring AI tools as well as application infrastructure.

SiliconANGLE reported that Groundcover said the new capabilities are available now. The same report said Groundcover has raised $160 million to date, including a $100 million Series C led by One Peak in July 2026, and that it made its first acquisition in September 2026 by buying Kubernetes resource optimization startup Wand Cloud Ltd.

Taken together, Groundcover is pushing observability toward an execution layer: investigate in one place, hand off to coding agents when needed, and keep the workflow inside the customer’s cloud account. For teams already using Claude or Codex in production support and engineering work, the pitch is less about a new dashboard and more about shortening the path from incident detection to remediation.