+ P Observe, evaluate, and improve your Google ADK agents
Capture every event flowing through Google ADK's own Event Loop — the same message format the Runner uses between your agents, the LLM, and tools — as structured trace data.
What Prefactor records from Google ADK
Google ADK + Prefactor
Observe for Google ADK
Prefactor observes your Google ADK agents in real time — every LLM call, tool invocation, and custom span captured as st
Open → EvaluateEvaluate for Google ADK
Prefactor evaluates your Google ADK agents — score outcome quality against the captured spans, track drift by comparing
Open → ObserveAct for Google ADK
Prefactor acts on your Google ADK agents at runtime — block, throttle, sandbox, or escalate a tool call or data access b
Open →How the Google ADK integration works
- ADK's Runner drives an Event Loop where Events are the standard message format between the UI, the Runner, your agents, the LLM, and tools — Prefactor spans attach at the same event boundaries, not a parallel structure alongside them.
- ADK's own before_agent_callback/after_agent_callback, before_model_callback/after_model_callback, and before_tool_callback/after_tool_callback hooks are the real interception points — the before_* callbacks are what a runtime policy runs through to inspect or block a call before ADK executes it.
- Beyond auto-captured spans, use withSpan to record any custom step you define — an API call, a quality check, a business action.
Google ADK integration FAQ
Do I need a dedicated package for Google ADK?
You can instrument Google ADK today with the framework-agnostic prefactor-core SDK; a dedicated package can be added on request.
What does Prefactor capture from Google ADK?
Prefactor records agent-tree steps, tool invocations and LLM calls as structured, timestamped spans — so every Google ADK run is captured as trace data you can reconstruct, search and export end to end.
Does Prefactor add latency or change how Google ADK runs?
No. Observability capture is designed to stay off your agent's critical path, so it doesn't alter your Google ADK logic or your users' responses. The only part that acts inline is the optional runtime guardrails you enable per agent — by design, so a high-risk or low-confidence action can be held for human approval before it executes.
Can I evaluate agents built with Google ADK and catch regressions?
Yes. Once runs are captured, eval suites score quality and groundedness on real traffic, drift detection flags behaviour changes after deployment, and versioned eval history catches regressions before they ship — the observe → evaluate → improve loop applied to your Google ADK agents.
Related guides
See it on your Google ADK agents
Book a 15-minute setup and our team gets you tracing in production.