Humans get plain-language orientation while agents get explicit routes, schemas, examples, and public metadata.
Inspect machine routesPhase 2.46 / v1.5.65
LocalEndpoint in plain language
LocalEndpoint is a public discovery and validation layer for local-first AI connections. It gives people and agents one place to understand what can be connected, what evidence exists, and what remains intentionally disabled.
Connection model
LocalEndpoint separates public knowledge from private authority.
The website publishes the vocabulary, validation rules, route metadata, and evidence shapes. Local machines keep runtime access, secrets, and approval decisions local.
Signal decoder
Three questions should explain the whole system.
The overview page turns LocalEndpoint into a readable decision surface: what the website is, what a visitor can safely do here, and where authority leaves the public web.
Use browser-local validation and public artifacts without uploading private endpoint files or exposing device state.
Try the validatorRuntime authority, credentials, command execution, and endpoint decisions stay outside the public website.
Review Desktop pathSignal relay
One local endpoint story, four trust signals.
The overview keeps each audience in its lane: visitors understand the boundary, developers inspect contracts, AI agents read public context, and Desktop keeps approval on the device.
Plain language shows what this site can explain, validate, and hand off before anyone downloads or approves.
Schemas, route indexes, examples, and quality gates make integration behavior inspectable.
Machine-readable routes provide context while public metadata still carries no runtime permission.
Private decisions remain on the device, where the person can review evidence and approve the handoff.
Trust topology
One public interface. Three separated authority zones.
LocalEndpoint is impressive because it is constrained: public discovery explains the system, browser validation checks artifacts locally, and Desktop approval owns private runtime decisions.
Readable by humans, crawlers, and agents. Never a remote control plane.
Runs in the visitor's browser without upload intake or localhost probing.
Private runtime authority stays on the device and remains human-visible.
Public-to-local handoff rail
The website prepares the request. The device decides.
LocalEndpoint turns the messy question of local AI connection into a sequence people can inspect: discover the public contract, validate the shape, export evidence, and hand authority to the local approval surface only when a person is ready.
- 01Discover
Read public identity, route metadata, docs, and machine guidance.
- 02Validate
Check manifest shape in the browser with no upload intake.
- 03Evidence
Export a reviewable receipt so claims become portable.
- 04Approve
Move decisions to Desktop, where local authority belongs.
Operating boundary
Public clarity, local authority.
This public site is static metadata and does not dispatch desktop commands, probe localhost, upload files, collect telemetry, request credentials, or claim runtime safety certification.