FOR AGENTS & DEVELOPERS
What this site exposes
This site runs a real, protocol-compliant agent surface — not descriptive text. Agents can read four resources (bio, case-studies, engagements, contact-card) and call two tools (evaluate_client_needs and schedule_consultation— which emails a real consultation request to Tony's inbox). All routes serve structured JSON; errors follow a single { error: { code, message, resolution } } shape.
Endpoints
Machine-readable spec of the entire API surface (OpenAPI 3.1). Start here if you are building an integration.
Self-describing manifest listing every resource and tool with its URL, method and body shape.
Protocol-compliant MCP server — the same surface as the REST API, for MCP-aware clients.
MCP-ecosystem server manifest advertising name, version and endpoint.
Plain-markdown site index for LLM agents, per the llms.txt convention.
Quick start
# Read the manifest
curl https://tonyhughesdesign.com/api/agent
# Fetch the case studies resource
curl https://tonyhughesdesign.com/api/agent/resources/case-studies
# Evaluate client needs (tool)
curl -X POST https://tonyhughesdesign.com/api/agent/tools/evaluate-client-needs \
-H "Content-Type: application/json" \
-d '{"challenges":"pitch not landing with investors","technology_sector":"fusion energy"}'
# Hand off a real consultation request (tool — emails Tony)
curl -X POST https://tonyhughesdesign.com/api/agent/tools/schedule-consultation \
-H "Content-Type: application/json" \
-d '{"client_name":"Jane Doe","client_email":"[email protected]",
"project_context":"decentralised AI platform pre-seed"}'Content negotiation & markdown
The homepage honours Accept negotiation: request it with Accept: text/markdownand you receive the page's content as markdown rather than the HTML shell (see acceptmarkdown.com). Static pages are also available as plain text via the sitemap and llms.txt.