Hermai offers open‑source, stable APIs that replace custom web scrapers by providing structured, site‑specific interfaces for frequently accessed sites such as government records, business filings, procurement portals, and market data sources. Users install a single skill that lets their agents read site schemas, query the site, and receive clean, token‑efficient data without HTML parsing. The platform also allows contributors to add new site schemas on demand.
Funding
Funding not disclosed
Founders
Product
Problem
Developers and automation agents often rely on custom web scrapers to extract data from sites that lack public APIs. These scrapers are fragile, break when page layouts change, and require ongoing maintenance, leading to unreliable data pipelines and high token costs for parsing raw HTML.
Solution
Hermai provides an open catalog of site‑specific APIs that replace fragile scrapers with stable, schema‑driven interfaces. By reading each site's data schema and handling HTTP calls automatically, the platform returns clean, structured data directly from sources such as government records, business filings, procurement portals, and live market feeds. The service eliminates the need for HTML parsing, reduces token consumption, and ensures consistent results even as site layouts evolve. Developers can extend coverage by adding new site schemas through the Hermai skill for Claude Code, Cursor, or Codex, enabling agents to query any supported site with a simple API call.
Target Audience
Primary users are developers, data engineers, and AI agent teams that need reliable, structured data from websites lacking official APIs, particularly in government, business, procurement, and market data domains.
Features
- Open, community‑maintained catalog of site‑specific APIs for sites without native public endpoints
- Automatic schema detection and request handling that returns structured JSON data instead of raw HTML
- Token‑efficient responses that minimize parsing overhead for LLM agents and downstream workflows
- CLI skill integration (`hermai` skill) for Claude Code, Cursor, and Codex, allowing agents to add or update site schemas on demand
- Benchmarking against generic scraper tools (e.g., Firecrawl, WebFetch) demonstrating higher data completeness and lower token usage
- Simple contribution workflow for adding new site schemas, ensuring the catalog stays up‑to‑date with minimal effort