Telemetry
AI operations

agent_tool_call_completed event schema

A bounded terminal tool-call event for agent reliability analysis. Keep arguments and results in the specialist trace system and retain only approved categories and correlation identifiers.

Reviewed by the Telemetry product team on . Event grain, ownership, field types, privacy classification, example payload, and validation. Review standards and ownership

Question this contract answers

Which tools fail, retry, loop, or add latency before an agent reaches a useful outcome?

Fields
11
Required
10
Contract evidence path

agent_tool_call_completed: from owned boundary to trusted query

Treat agent_tool_call_completed as a durable analytical contract: the owner emits one documented grain, a fixture proves the fields, and a reviewed query answers the stated question.

  1. 1

    Outcome becomes final

    Agent runtime or tool-execution adapter emits only after the tool returns, times out, is denied, or fails for the current attempt.

  2. 2

    Contract is bounded

    10 required fields preserve the declared grain: One completed tool-call attempt within an agent run.

  3. 3

    Fixture is verified

    Check types, UTC time, alternate outcomes, idempotency, and every pseudonymous or review-classified field.

  4. 4

    Question is answered

    Which tools fail, retry, loop, or add latency before an agent reaches a useful outcome?

Grain

One completed tool-call attempt within an agent run.

Owner

Agent runtime or tool-execution adapter

Emit when

After the tool returns, times out, is denied, or fails for the current attempt.

Field contract

Typed fields with explicit privacy boundaries

Keep existing field names and types stable after production queries depend on them. Optional context should remain bounded, documented, and justified by a specific decision.

FieldTypeRequiredPrivacyMeaning
timestamp_utctimestampyesnon-sensitiveUTC time at the outcome boundary.
event_idstringyesnon-sensitiveStable unique identifier used for deduplication.
releasestringyesnon-sensitiveApplication or service version that emitted the event.
account_idstringyespseudonymousStable internal account identifier, never an email or name.
run_idstringyespseudonymousApproved identifier for the owning agent run.
tool_call_idstringyespseudonymousStable identifier for this logical tool call.
tool_namestringyesnon-sensitiveAllowlisted low-cardinality tool name.
statusstringyesnon-sensitiveSuccess, failed, timed_out, or denied.
attemptnumberyesnon-sensitiveOne-based attempt number for the logical call.
duration_msnumberyesnon-sensitiveTool-call duration in milliseconds.
error_typestringnonon-sensitiveControlled failure category without exception text.

Synthetic JSON event

{
  "timestamp_utc": "2026-07-29T10:04:11Z",
  "event_id": "evt_tool_01",
  "account_id": "acct_8f31",
  "release": "2026.07.3",
  "run_id": "run_43fc",
  "tool_call_id": "call_017",
  "tool_name": "customer_lookup",
  "status": "success",
  "attempt": 1,
  "duration_ms": 284
}

Privacy review

Review identifiers before ingestion

This example uses synthetic identifiers. Pseudonymous values can still be personal data, and review fields can expose business or provider context. Apply your own consent, retention, access, residency, and deletion requirements.

  • account_id: pseudonymous
  • run_id: pseudonymous
  • tool_call_id: pseudonymous

Validation checklist

Prove the contract before building a dashboard

  • Send one known agent_tool_call_completed fixture after the documented outcome boundary.
  • Verify all 10 required fields arrive with the documented types.
  • Retry the same event identifier and confirm the chosen deduplication behavior.
  • Send a controlled failure or alternate outcome when the workflow supports one.
  • Run the related SQL over a fixed window and reconcile the result to the fixture.

Common mistakes

Keep one row equal to one durable outcome

  • Emitting agent_tool_call_completed before agent runtime or tool-execution adapter knows the final outcome.
  • Mixing several grains in one table, which makes counts and rates ambiguous.
  • Replacing controlled categories with raw URLs, payloads, prompts, or error text.
  • Changing a field type in place after saved queries and dashboards depend on it.
  • Adding identifiers without a documented investigation, access, and retention need.

Use the contract

Query and operationalize the event

Related contracts

Send a fixture before production traffic

Create a free API key, send the synthetic event, and inspect the inferred table before connecting a live workflow.

Test this schema