ReplayAutonomous app testing for teams who ship at AI-speed
Promoted
Manifest
Turn any webpage into an action manifest for AI agents
178 followers
Turn any webpage into an action manifest for AI agents
178 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.
Finally a tool that captures the "why" behind clickable elements, not just the selectors. The requires dependency mapping is genuinely useful and saved me a ton of trial and error on a flaky login flow.
Manifest
@utkufqkz
Thanks, Utku — that "why, not just the selector" framing is exactly what I was going for, since a bare selector tells you an element exists but not what has to happen before it's actually usable. Glad requires cut down the trial and error on a flaky login flow — that's a classic case where the ordering isn't obvious until something silently fails.
the requires field is such a smart touch, encoding dependencies directly into the map saves so much back-and-forth between agent and page. clean SDK and MCP server setup too.
Manifest
@ferhattkge
Thanks, Ferhat — that back-and-forth is exactly what I wanted to cut out. Without the dependency info baked in, the agent's stuck poking at the page to figure out what's actually available, which burns cycles it shouldn't need to. Glad the SDK and MCP setup are landing clean too — appreciate you calling that out specifically.
the requires field for element dependencies is genuinely useful, something I've hacked together badly on past projects. excited to try it with an MCP server setup
Manifest
@hanmtanrkanwyt
Thanks, Hanım — "hacked together badly" is a pretty universal experience with this problem, which is exactly why I wanted it baked into the response instead of something everyone has to rebuild themselves. Hope the MCP server setup makes it an easy first try — let me know how it goes.
The requires field encoding dependencies between elements is the interesting part — "select a plan before this button is clickable" is exactly the context accessibility trees drop, and it's where most browser agents burn their retries. Question: how do you handle drift? If the site ships a redesign, does the manifest regenerate per-call (so it's always fresh but you pay latency), or is there caching with some invalidation heuristic? That trade-off decides whether I'd trust it in a scheduled job.
Action manifests are a useful primitive for agent workflows. The part I would want to see emphasized is preconditions and permissions: what the page action is allowed to do, what success means, and what evidence gets handed back after the run.
Manifest
The dependency graph between elements is genuinely useful for agent builders, something I've been piecing together by hand for months. One thing that would save a ton of debugging time: let me simulate an action through the same API and tell me whether the requires chain actually cleared or if something is silently blocking it. Right now I can see the map, but I can't trust it until I try it.
Manifest
The requires field for encoding element dependencies is genuinely clever. It's the kind of context that usually takes teams weeks of brittle DOM mapping to figure out, baked right into the response.
Manifest
@uurk6oe
Thanks, Uğur — that's exactly the gap I was aiming at. Teams end up hand-mapping that dependency logic per site, and it's brittle the moment the DOM shifts, so baking it into the response instead of leaving it as tribal knowledge in someone's scraper script was the whole point. Appreciate you naming why it matters, not just that it exists.