Connect your AI tools to this site

MCP server

https://liamogrady.dev/mcp

Model Context Protocol over Streamable HTTP (current spec, with 2025-era clients supported). No authentication, rate-limited, read-only. Tools: get_profile, get_experience, list_projects, get_certifications, search_profile, assess_role_fit and get_contact_options. Resources: liam://profile.md, liam://profile.json. Prompts: introduce_liam, assess_fit.

Connect

# Claude Code
claude mcp add --transport http liam https://liamogrady.dev/mcp

# VS Code (or use the one-click link on the home page)
code --add-mcp '{"name":"liam-ogrady","type":"http","url":"https://liamogrady.dev/mcp"}'

# Copilot Studio
Tools → Add a tool → New tool → Model Context Protocol → URL above → Authentication: None

# Microsoft 365 Copilot (Agents Toolkit)
Declarative agent → Add an action → Start with an MCP server → URL above

WebMCP

When a browser agent supports WebMCP (document.modelContext), the home page registers tools for it directly: search_profile, get_section, navigate and open_agent. They run in the visitor's browser and never send data anywhere else.

Plain files

• /llms.txt: a short map of the site for language models
• /llms-full.txt: the whole profile in Markdown
• /cv: printable HTML CV with schema.org JSON-LD
• curl liamogrady.dev: an ANSI CV for terminals

The chat API

POST /api/agent streams Server-Sent Events: meta, status, tool, tool_result, ui, card, text, done. UI events let the chat scroll and highlight this page. It only uses the public profile, does not discuss clients, and falls back to plain search when the model is unavailable or the daily budget is spent.