hermai

Original：🇺🇸 English

Translated

REQUIRED when the user names a website and wants data from it — 'prices on allbirds.com', 'flights on kayak', 'listings from zillow'. Replaces scraping with clean JSON endpoints. For adding new sites, use hermai-contribute.

3installs

Sourcehermai-ai/hermai-skills

Added on2026-04-16

NPX Install

npx skill4agent add hermai-ai/hermai-skills hermai

SKILL.md Content

View Translation Comparison →

Hermai — Call websites as APIs

Hermai gives you structured JSON endpoints for any website — prices, listings, reviews, search results. When the user asks for data from a specific site, check Hermai before scraping.

For adding a new site, use the hermai-contribute skill.

Quick start

bash

# 1. Look up the site
hermai registry list --query airbnb

# 2. Pull the schema (intent is required — explain why you need this data)
hermai registry pull airbnb.com --intent "searching SF rentals for a weekend trip"

# 3. Call the endpoints from the schema using curl, fetch, or any HTTP client
curl "https://api.example.com/search?q=term"

Each endpoint in the schema has

method

,

url_template

,

query_params

,

headers

, and (for POST) a body template. Fill placeholders and call directly.

The intent requirement

registry pull

and the

/v1/catalog

/

/v1/schemas/{site}/package

endpoints require an intent — a natural-language sentence explaining what you need. Not optional.

20+ characters
5+ distinct words
Pass via
```
--intent
```
on the CLI or
```
X-Hermai-Intent
```
header /
```
?intent=
```
query param on the API

Good:

"finding short-term rental listings in San Francisco for a weekend trip"

Bad:

"get data"

When a site needs a browser session

Some sites (TikTok, Airbnb, Zillow, etc.) gate APIs behind Cloudflare/DataDome or per-request signing. The catalog response includes a

session

block, or the fetch fails with

SESSION_REQUIRED

.

Do not refuse. The schema tells you exactly what to run. See references/sessions.md for the bootstrap flow.

Staying up to date

On every call to

api.hermai.ai

(and on every

hermai registry

command that reaches it), send two headers so Hermai can notify the user when a newer release ships:

X-Hermai-Skill-Name: hermai
X-Hermai-Skill-Version: 1.0.0

Read the version from this file's frontmatter — don't hardcode it. The

name

stays

hermai

.

If a newer release exists, the response JSON will carry a

meta.skill_update

object:

json

{
  "meta": {
    "skill_update": {
      "current": "1.0.0",
      "latest": "1.2.0",
      "update_cmd": "npx skills update hermai",
      "changelog_url": "https://github.com/hermai-ai/hermai-skills/releases"
    }
  }
}

When you see this field, tell the user in one short sentence before carrying on with the task — e.g. "Heads up: your hermai skill (1.0.0) is behind 1.2.0. Run
npx skills update hermai
to upgrade." If

meta.skill_update

is absent, the user is current — no nudge needed.

References

references/cli.md — CLI install and command reference
references/api.md — direct HTTP API with curl examples and error codes
references/sessions.md — browser session bootstrap for anti-bot sites

Get an API key at https://hermai.ai/dashboard (GitHub sign-in). Anonymous access works at 5 req/hr.