Manifest
@noma-py
Gives AI agents a structured action manifest for any webpage — what's clickable, fillable, and submittable, so agents can act on a page instead of just reading it.
Overview
What is Manifest?
Manifest is a service that turns any webpage into a structured action manifest for AI agents. Instead of returning raw HTML or a screenshot, it returns JSON describing clickable, fillable, and submittable elements on the page.
How to use Manifest?
Get an API key from app.manifest.omfang.io, then call the get_manifest tool with a URL. The server returns a structured manifest of actions. Manifest also works as a Python SDK (pip install manifest-api) and has a LangChain tool integration.
Key features of Manifest?
- Returns click, fill, and submit actions, not just page content
- Detects required fields, input types, and disabled elements
- Caches manifests for near-instant repeat calls
- Works as MCP server, Python SDK, or LangChain tool
Use cases of Manifest?
- Let an AI agent fill and submit web forms without hardcoded selectors
- Navigate sites with frequently changing UIs reliably
- Build browser-automation agents that reason about available actions
FAQ from Manifest
What’s the difference between Manifest and a browser automation tool like Browserbase?
Browserbase gives agent access to a browser; Manifest gives structured understanding of actions on a page. They are complementary, not competing.
Does Manifest require an API key?
Yes, get one at app.manifest.omfang.io. A free tier is available.
What does the response look like?
A JSON object with the page's current state, a list of actions (type, label, description, required flag), and navigation links.
