Connect your AI to the commons
NTLSN operates a hosted Model Context Protocol (MCP) server. Any MCP-capable AI client — Claude Code, Claude Desktop, or any other compliant tool — can query the national teaching-and-learning commons directly: upcoming sector events, the rescued ALTC/OLT archive, and the Learning & Teaching landscape across 42 Australian universities. It is read-only, uses only the open data already published on this site, and requires no account or key.
What is the Model Context Protocol?
The Model Context Protocol is an open standard, originally released by Anthropic in 2024, that lets AI assistants call external tools and data sources in a structured, auditable way. Where a chatbot ordinarily answers from its training data — which may be stale or wrong — an MCP-connected assistant can put a live, verifiable question to a real data source and cite what comes back. For an academic audience the useful analogy is interlibrary loan rather than memory: the assistant requests the record, receives it, and works from the actual document. NTLSN publishes its commons data through MCP so that the sector's increasingly AI-mediated workflows draw on curated, current, openly published information rather than approximation.
The endpoint
https://ntlsn.com/.netlify/functions/mcpThe server speaks JSON-RPC 2.0 over HTTPS. A POST carries MCP requests (initialize, tools/list, tools/call, ping); a plain GET returns a machine- and human-readable description of the server and its tools, so the URL is self-documenting. Responses are JSON. Cross-origin requests are permitted.
Try it now from a terminal
These three commands work today, unchanged, from any machine with curl. They are the ground truth for this server: if a particular client's configuration ever gives trouble, what these commands return is what the server actually does.
1. Initialise a session
curl -s -X POST https://ntlsn.com/.netlify/functions/mcp \
-H 'Content-Type: application/json' \
-d '{"jsonrpc":"2.0","id":1,"method":"initialize","params":{"protocolVersion":"2024-11-05","capabilities":{},"clientInfo":{"name":"curl","version":"0.0.1"}}}'Returns the server identity: {"serverInfo":{"name":"ntlsn","version":"1.0.0"}} with protocol version 2024-11-05.
2. List the available tools
curl -s -X POST https://ntlsn.com/.netlify/functions/mcp \
-H 'Content-Type: application/json' \
-d '{"jsonrpc":"2.0","id":2,"method":"tools/list"}'Returns the three tools documented below, with their full input schemas.
3. Call a tool — the next two sector events
curl -s -X POST https://ntlsn.com/.netlify/functions/mcp \
-H 'Content-Type: application/json' \
-d '{"jsonrpc":"2.0","id":3,"method":"tools/call","params":{"name":"upcoming_events","arguments":{"limit":2}}}'At the time of writing this returns HERDSA Conference 2026 and the ACODE Benchmarking Summit 2026, drawn live from the same curated events dataset that powers this site. Your result will reflect the calendar on the day you run it.
Connect an AI client
Claude Code (terminal)
Claude Code supports remote HTTP MCP servers directly. One command:
claude mcp add --transport http ntlsn https://ntlsn.com/.netlify/functions/mcpOr, for a project-scoped setup, add this to the project's .mcp.json:
{
"mcpServers": {
"ntlsn": {
"type": "http",
"url": "https://ntlsn.com/.netlify/functions/mcp"
}
}
}Claude Desktop and claude.ai
Claude Desktop's configuration file (claude_desktop_config.json) accepts only locally launched servers — it does not take a remote URL directly. Two documented routes work for a remote server such as this one:
- Custom connector (no installation). In Claude's settings, open Connectors, choose "Add custom connector", and paste the endpoint URL above. No authentication is required — leave the OAuth fields empty.
- The mcp-remote bridge (requires Node.js). Add the following to
claude_desktop_config.json; it launches a small community-maintained proxy that speaks HTTP to this server on the client's behalf:{ "mcpServers": { "ntlsn": { "command": "npx", "args": ["mcp-remote", "https://ntlsn.com/.netlify/functions/mcp"] } } }
These client configurations follow the vendors' current published documentation. Every claim on this page was verified against the live endpoint on 5 July 2026; client software changes on its own schedule. If a client-side step fails, the curl commands above remain the definitive test of the server.
Any other MCP client
Any client that supports remote MCP servers over HTTP can use the endpoint URL as-is. The server implements initialize, tools/list, tools/call, and ping, accepts standard MCP notifications, and returns conventional JSON-RPC errors for anything else.
What the server offers
Three read-only tools, each drawing on data openly published on this site.
upcoming_events
Upcoming Australian higher-education teaching-and-learning events — conferences, showcases, workshops and webinars — from NTLSN's curated events dataset (each record carries a verified flag).
| Argument | Type | Required | Notes |
|---|---|---|---|
| uni | string | no | Filter to one university by short id, e.g. "usq" |
| limit | integer | no | Maximum results; default 10 |
search_archive
Searches the rescued archive of 1,431 ALTC- and OLT-funded teaching-scholarship works (1994–2025) by title or author. This is the national grant-funded teaching-scholarship record that NTLSN preserves and republishes as open data.
| Argument | Type | Required | Notes |
|---|---|---|---|
| query | string | yes | Title or author search terms |
| limit | integer | no | Maximum results; default 15 |
universities
Returns the 42 Australian universities in the commons, each with its central Learning & Teaching team page and alliance-group membership. Takes no arguments.
| Argument | Type | Required | Notes |
|---|---|---|---|
| — | — | — | No arguments |
Things worth asking, once connected
- "What teaching and learning conferences are coming up in Australia in the next two months?"
- "Search the ALTC/OLT archive for work on peer review of teaching."
- "Which universities' Learning & Teaching teams should I look at, and where are their pages?"
What this server does — and does not do
- Read-only. The server answers questions. It cannot create, modify, or delete anything, and it accepts no uploads.
- Public data only. Every response is drawn from the two open datasets already published on this site —
/data/events.jsonand/data/ltr.json— plus the public university directory. Nothing is available through MCP that is not available in a browser. - No accounts, no keys. There is no sign-up, no token, and no way to identify yourself to the server even if you wanted to.
- Nothing stored. The server keeps no database and writes no application-level record of your queries. Requests pass through Netlify's hosting platform, which maintains routine operational logs as any host does.
- Fair use. The endpoint is subject to the hosting platform's standard rate limits. It exists for genuine scholarly and educational use; please build considerate clients.
The data behind the tools
Both source datasets are published open access: /data/events.json (the curated national events calendar) and /data/ltr.json (the ALTC/OLT archive metadata). A formal licence statement for these datasets is in progress; until it is published, they are offered for scholarly and educational use, and we ask that you contact us before large-scale redistribution.
Independence
NTLSN is independent of, and not affiliated with, Anthropic — originators of the Model Context Protocol — or any other AI vendor. MCP is an open standard, and this server works with any compliant client; the configuration examples above name specific products solely because they are widely used, and imply no endorsement in either direction. The mcp-remote bridge referenced above is community-maintained software, likewise unaffiliated with NTLSN. NTLSN receives nothing from any vendor named on this page.
Server status: a plain GET to the endpoint returns its self-description — the simplest health check there is.
← Back to NTLSN · Built to connect → · Privacy → · hello@ntlsn.com