Instrument the bounded research agent

Instrument the bounded research agent over a fixture KB so a run records enough structure to debug offline: counts, latency if relevant, pass/fail of step count ≤ max, and a stable stage name. Redact secrets and raw credentials from every event.

const event={stage:'research-agent', steps:2, citations:1, abstained:false};
console.log(JSON.stringify(event));

Expected evidence: agent telemetry. Prefer JSON or structured text you can grep in CI over prose logs for agents-in-code.

Turn one historical failure—especially infinite tool loop—into a test that fails if the signal disappears for the bounded research agent. Observability without a failing test is optional decoration; observability with a test is part of the agents-in-code artifact.

Draft a three-field event for the bounded research agent: stage, ok, and one domain field derived from step count ≤ max; citation coverage; abstention on empty search. Add fixture_id or docs_version when content can change. Explicitly list fields that must never appear (tokens, passwords, raw prompts) because agent inventing tools outside the registry mid-run is in scope for this lab.

Wire one assertion that fails if the bounded research agent event is missing after a run. Observability that cannot fail a test will not survive contact with a busy agents-in-code repository.

Imagine a teammate opens only your event stream after a bad deploy. Could they tell whether fixture KB with 4 notes + max_steps=3 was wrong, whether infinite tool loop, or final answer with no citations after search returned, or whether agent inventing tools outside the registry mid-run slipped through? If not, rename fields until those three stories are distinguishable.

At the testing and observability stage for agents-in-code, the job is narrower than finishing a product demo. You are creating one progressive evidence piece about fixture KB with 4 notes + max_steps=3 that later pages inherit without redefining success. Keep that fixture small enough to inspect by hand, keep outputs copy-pasteable as text, and refuse to narrate this baseline as if it were a production SLA: single-shot LLM answer without search.

For this page specifically, success looks like a structured event schema locked by a test while still centering the user decision to search a local knowledge base then produce a cited answer within a step budget. If you cannot point to a file, command, or assertion that proves that for the bounded research agent, stay on this page instead of advancing.

Glossary: tool · Glossary: structured output · Cheatsheet: production ops signals

Previous · Next