Skip to content

ToolCallJudgeOptions

Matcher context accepted by ToolCallJudge().

await expect(result).toSatisfyJudge(ToolCallJudge(), {
expectedTools: ["lookupInvoice", "createRefund"],
});

optional expectedTools?: ToolCallJudgeExpectedTool[]


harness: any

Harness associated with this judge context.

JudgeContext.harness


input: any

Original eval input passed to the harness.

JudgeContext.input


metadata: Readonly<TMetadata>

Per-run expectations or configuration passed to run(input, { metadata }).

JudgeContext.metadata


output: any

App-facing output returned by the harness.

JudgeContext.output


run: HarnessRun<any>

Complete normalized harness run being judged.

JudgeContext.run


optional runJudge?: RunJudge

Runs the configured matcher, judge, or suite judge harness with run-scoped context.

JudgeContext.runJudge


session: NormalizedSession

Normalized transcript associated with the harness run.

JudgeContext.session


toolCalls: ToolCallRecord[]

Flattened tool calls observed in the normalized session.

JudgeContext.toolCalls