Show HN: Shotlist – Make your AI agent prove its work with real screenshots
Shotlist is a tool that requires AI agents to provide real screenshots as proof of their work, adding accountability and verifiability to automated tasks.
Background
- Shotlist is an open-source tool that automatically captures screenshots as an AI agent performs tasks, creating a visual audit trail of its actions.
- It addresses the "black box" problem in AI automation: when an agent claims to have completed a task, there's often no way to verify what it actually did or whether it made mistakes.
- The tool works with popular AI agent frameworks and browser automation tools (like Playwright, Puppeteer, or browser-use), injecting screenshot capture at each step.
- This is part of a broader trend toward "observability" and "trustworthiness" in AI — as companies deploy autonomous agents for real work, they need accountability and debugging capabilities similar to what logging provides for traditional software.
- The project's GitHub repository makes it freely available, targeting developers building AI-powered automation workflows who need to verify agent behavior.