Files
deepagentsjs/evals/basic/index.test.ts
T
Hunter Lovell b30cdf9064 feat: add eval harness and initial eval suites (#309)
* Squashed '.agents/skills/langsmith-skills/' content from commit aec5e8f

git-subtree-dir: .agents/skills/langsmith-skills
git-subtree-split: aec5e8f059ff50b1cde655e50b27a25da7981e60

* feat: add eval harness and initial eval suites

* cr

* cr

* cr

* cr

* cr

* cr

* cr
2026-03-17 10:06:55 -07:00

47 lines
1.2 KiB
TypeScript

import * as ls from "langsmith/vitest";
import { expect } from "vitest";
import { getDefaultRunner } from "@deepagents/evals";
const runner = getDefaultRunner();
ls.describe(
"deepagents-js-basic",
() => {
ls.test(
"system prompt: custom system prompt",
{
inputs: { query: "what is your name" },
referenceOutputs: { expectedText: "Foo Bar" },
},
async ({ inputs }) => {
const result = await runner
.extend({ systemPrompt: "Your name is Foo Bar." })
.run({ query: inputs.query });
expect(result).toHaveFinalTextContaining("Foo Bar");
ls.logFeedback({
key: "agent_steps",
score: result.steps.length,
});
},
);
ls.test(
"avoid unnecessary tool calls",
{
inputs: { query: "What is 2+2? Answer with just the number." },
},
async ({ inputs }) => {
const result = await runner.run({ query: inputs.query });
expect(result).toHaveFinalTextContaining("4");
ls.logFeedback({
key: "agent_steps",
score: result.steps.length,
});
},
);
},
{ projectName: runner.name, upsert: true },
);