Catalog and hosted runtime.

Everfetch publishes factual metadata about publicly observable website endpoints and runs authenticated hosted fetches for customers. This page explains what we publish, what we execute, and what customers remain responsible for.

The catalog / runtime split

Everfetch hosts a catalog of schemas — structured descriptions of what public websites return. A schema might say “Booking.com has a hotel search endpoint that takes destination, check-in, check-out, and guests, and returns name, price, and review score.” That is a factual description of a public workflow.

When you run local tooling such as everfetch fetch from your own machine, your machine sends the request and you are the actor. When you call the hosted Story Ginie API with an API key, Everfetch executes the request from our managed infrastructure and returns the requested metadata or workflow result.

In both modes, customers choose what to request and must use Everfetch lawfully, with any permissions or contractual rights required for their use case.

Customer data commitments

• We do not train models on customer inputs or outputs.
• We do not claim ownership of customer inputs, outputs, prompts, URLs, or results.
• Hosted fetch requests are processed to provide, secure, debug, and bill the service.
• Our primary hosted API and fetch infrastructure currently runs in the United States.

Security controls

• Public website and API traffic is served over HTTPS/TLS.
• Data stored in our primary Google Cloud data stores is protected by Google Cloud's default encryption at rest.
• API keys are generated with cryptographic randomness, validated by hash, and can be revoked.
• If we become aware of a security incident affecting account or customer data, we will investigate and notify affected customers as required by law or contract.
• We can provide current security-review materials, subprocessor details, and certification status during vendor review.

Provenance and audit trail

Every catalog entry is traceable. We treat provenance as a feature, not an afterthought, because compliance and audit workflows depend on it:

• Each schema records its source, endpoint surface, and last-verified time.
• Catalog entries publish availability and last-verified metadata where it has been established; verification coverage varies by endpoint.
• Change history is retained so you can see when a source's shape last moved.
• Hosted responses are tied to the schema version that produced them, so a value can be traced back to where it came from.

What we publish

The public catalog shows high-level facts. Entries can include:

• Schema name, site, and public purpose
• Intent category, such as travel
• Endpoint count and safe endpoint previews
• Public availability and verification signals
• Higher-cost or additional-setup flags where relevant

What stays private

The full technical recipe lives only in pulled packages, on contributor and user machines. The public catalog may show safe examples and high-level endpoint names, but it does not publish the private package details:

• Private endpoint variants and execution recipes
• Request headers, payloads, and credentials
• Detailed validation rules and nested implementation shapes
• Full response type definitions
• Authentication flow details
• Selectors, parsing instructions, and site-specific handling

Runtime execution details are treated as operational details of the local client or hosted runtime and are not exposed in the public catalog.

Legal posture

We believe publishing factual interoperability metadata about public website endpoints, and accessing publicly available data, rests on well-established legal ground. This is general background, not legal advice. If you have questions about a specific schema or use case, email hello@everfetch.tech.

Related pages

• Takedown process: how site operators can request removal.
• Privacy policy: what we collect, what we don't.

Contact

For takedowns, partnership questions, security disclosures, or compliance requests, email hello@everfetch.tech. Every message is read by a person, and we aim to reply promptly.