Agent API

Test console for agentic business actions.

This page shows how an AI agent discovers agentready.ch, reads company profiles and runs approved actions. Contact requests are delivered when a recipient is configured; bookings can still require confirmation.

Reading

Reading is public

Manifest, profiles, sources, functions.

Contact

Contact is permissioned

Consent, sender identity and an allowed intent are required.

Sending

Execution follows company rules

Contact can be delivered directly; booking and stronger actions can still require confirmation.

Discovery

How an agent finds the functions

The entry point is the webMCP manifest. From there, links lead to OpenAPI, profiles and permitted actions.

/.well-known/webmcp.json

Discovery manifest for agents

/llms.txt

Site-level LLM orientation

/api/agent/openapi.json

OpenAPI schema for actions

/api/mcp

Streamable HTTP JSON-RPC MCP endpoint

/companies/{slug}/agent.json

Structured company profile

/companies/{slug}/agent.md

Markdown profile for LLM context

/api/agent/companies/{slug}/contact-request

Send a permissioned contact request

/api/agent/companies/{slug}/calendar-slots

Read available setup-call slots

/api/agent/companies/{slug}/calendar-booking

Prepare a confirmed calendar booking

Live test

Use a published company slug. For safe tests, send dry_run or preview_only; real contact requests are delivered when a recipient is configured.

Company Slug

No test run yet

ready

Choose Manifest, OpenAPI, agent.json or test contact.