Launched this week
Manifest
Turn any webpage into an action manifest for AI agents
180 followers
Turn any webpage into an action manifest for AI agents
180 followers
Manifest turns any webpage into a structured JSON map of what an AI agent can click, fill, and submit. One API call, no fragile selectors. Every action comes with resolved CSS/role locators, plus a requires field that encodes dependencies between elements (e.g. "select a plan before this button is clickable"). That's context aria snapshots don't give you. Python SDK, LangChain support, and an MCP server included. Built for anyone shipping browser agents.
Manifest
Update for everyone who commented on caching and repeat-call costs (a few of you flagged this independently, and @noctis06 specifically proposed the shape of it) — we shipped a /fingerprint endpoint.
It renders the page and returns a hash of the interactive elements, skipping the LLM extraction entirely. Same hash means nothing meaningful changed, so your agent can reuse its last manifest instead of paying for a fresh one every loop. Different hash means something real changed and it's worth re-extracting.
Live now in the API and in manifest-api 0.2.0 on PyPI. This one came directly out of this comment section, so — thank you. Let us know how it holds up in practice.
finally something that gets at the real pain point. the requires field encoding dependencies is genuinely clever, going to wire this into our langchain setup this week and see if it holds up on a few gnarly sites.
Manifest
@minekld0
Thanks, Mine — glad it's landing on the actual pain point rather than the surface-level stuff. The LangChain integration should make wiring it in pretty straightforward, but I'd genuinely like to hear how it holds up on the gnarly sites — that's exactly the kind of real-world stress test that surfaces where requires breaks down. Let me know how it goes.
This is really interesting. Can users choose which actions the agent can take on its own and which ones need approval first?