Skip to main content

Overview

Connectors let Warrnagen — Warrn’s AI agent — interact with external tools during incident investigation and response. Each connector exposes a set of tools that Warrnagen can call, with configurable access levels per tool. Connectors use the Model Context Protocol (MCP) to communicate with external services.

Available Connectors

GitHub

Access repositories, issues, pull requests, and code search.

Grafana

Query dashboards, alerts, and metrics for observability.

Coralogix

Search logs, traces, and alerts across your stack.

Custom MCP

Connect any MCP-compatible server.

How Connectors Work

  1. Install a connector from Settings → Connectors.
  2. Authenticate — provide API keys, or link to an existing integration (e.g., GitHub OAuth).
  3. Configure tool access — set each tool to Allow, Needs Approval, or Blocked.
  4. Bind to services — optionally map connector resources (repos, dashboards) to Warrn services.

Tool Access Levels

Each tool exposed by a connector can be set to one of three access levels:
LevelBehavior
AllowWarrnagen can call this tool automatically during investigations.
Needs ApprovalWarrnagen must request approval before calling. Default for most tools.
BlockedTool is completely disabled.

Service Bindings

Bindings map connector resources to Warrn services. For example, you can bind a GitHub repository to a service so Warrnagen knows which repo to search when investigating alerts for that service.

Custom MCP Servers

You can connect any MCP-compatible server by providing its URL. This lets you integrate internal tools, custom APIs, or any service that implements the MCP protocol.

Best Practices

  1. Start with Needs Approval for write operations — let the team build trust before allowing autonomous writes.
  2. Bind connectors to services so Warrnagen has the right context during investigations.
  3. Use the built-in connectors when available — they handle auth and tool discovery automatically.
  4. Review tool access levels periodically as your team’s comfort with AI automation evolves.