@promptsnap/sdk

Semantic snapshot testing for LLM apps. Wrap any LLM call with checkpoint() to capture its output, commit the snapshot to your repo, and get a PR comment describing exactly what changed.

Install

npm add -D @promptsnap/sdk @promptsnap/cli

Quick start

import { checkpoint } from '@promptsnap/sdk';
import OpenAI from 'openai';

const openai = new OpenAI();

export async function answerQuestion(question: string) {
  return checkpoint('qa/answer', async () => {
    const r = await openai.chat.completions.create({
      model: 'gpt-4o',
      messages: [{ role: 'user', content: question }],
      temperature: 0,
    });
    return r.choices[0]?.message?.content ?? '';
  }, { input: { question } });
}

Vitest — add one line to vitest.config.ts:

test: { setupFiles: ['@promptsnap/sdk/vitest'] }

Jest — add to jest.config.ts:

setupFilesAfterEnv: ['@promptsnap/sdk/jest']

Capture baselines

PROMPTSNAP_MODE=record npx vitest run
git add __llm_snapshots__ && git commit -m "chore: baselines"

Exports

| Path | Contents | |---|---| | @promptsnap/sdk | checkpoint(), determinism helpers, runtime mode | | @promptsnap/sdk/vitest | Vitest afterEach integration | | @promptsnap/sdk/jest | Jest afterEach integration | | @promptsnap/sdk/internal | Storage helpers (used by CLI) |

Documentation

→ docs.promptsnap.live