Investigation

An Investigation captures hypotheses, data snapshots, decisions, and outcomes so the entire on-call team can follow (or replay) the debugging story.

Runbook

A version-controlled, step-by-step workflow that encodes how to diagnose or fix a recurring class of issues. Built from Tasks + Variables + Markdown, and triggerable automatically (via alert rules) or on demand

Task

The smallest executable unit inside a runbook (or used ad-hoc). Examples: run kubectl get pods, query Datadog metrics, delete a stuck pod, request human approval. Tasks run sequentially or conditionally to form the full procedure.

Variables

Parameter placeholders (e.g., {{namespace}}, {{service_name}}) that make tasks and runbooks reusable across environments. Values are supplied at run time—by the alert payload, user input, or defaults.

Context

All supplemental knowledge the AI can reference: uploaded docs, imported Confluence pages, the Services & Infra catalog, previous investigations, etc. Rich context lets DrDroid produce cited, environment-aware diagnoses.

Proxy Services

Secure relay endpoints (managed in Settings → Proxy Service Tokens) that let the SaaS agent run commands inside your private network without opening inbound ports. Think of them as one-way “valves” for safe, audited execution.