API

Ask the ecosystem.
Programmatically.

Everything Canton Open Intelligence shows on screen is available to you as well — the registry, the Signal Feed, the analyst. No key, no signup, no rate-limit games. Just ask politely and cite the source when you publish.

MCP server

Four tools, one address.

Point any MCP client at https://canton.w3ai.io/mcp and your agent can search the Canton ecosystem the way our analyst does.

01

search_ecosystem

Keyword search across registry companies, their scraped homepages and GitHub summaries, canton.network pages, the Canton Deep Dive white paper and news.

query · kind · limit
02

get_project

Full profile for one company: description, categories, homepage content, GitHub organisations and top repositories, social links.

project
03

list_projects

List registry companies (id, name, categories), optionally filtered by category text.

category
04

ecosystem_stats

Headline counts for the snapshot: projects, GitHub coverage, Canton pages, news and category breakdown.

—
Endpoints

Plain HTTP, plain JSON.

Base address: https://canton.w3ai.io. Every one of these is public by design — they only ever return information we already publish.

GET/api/public/feed-items

The whole Signal Feed as JSON: id, url, title, description, host, source_type, project_ids, published_at, fetched_at, meta — plus the projects with GitHub coverage. Cached for 30 seconds, open to any origin.

POST/api/public/analyst-chat

Ask the research analyst. Body: { messages: [{ role, content }], page? }. Streams back a cited answer.

POST/api/public/canton-article

Rewritten version of one story, in our own words, facts and figures intact. Body: { url } — any HTTPS article address. Saved after the first rewrite, so later reads are instant.

POST/api/public/canton-feed

Scan a batch of company websites for new stories. Body: { cursor, sites: [{ id, name, website }] }. This is what the dashboard's Refresh feed button calls.

POST/api/public/research-search

Live web search with sources. Body: { question }.

POST/api/public/research

Streamed research on one project, with sources. Body: { projectId, … }. Same-origin only.

POST/api/public/research-suggest

Three suggested research prompts for a project. Body: { projectId, project }. Same-origin only.

curl -s https://canton.w3ai.io/api/public/feed-items | head -c 400 curl -s -X POST https://canton.w3ai.io/api/public/analyst-chat \ -H 'content-type: application/json' \ -d '{"messages":[{"role":"user","content":"Who is building custody on Canton?"}]}'
Ground rules
01

No key needed

These reads are open. The two scheduled jobs that keep the feed fresh, and the AI proxy the dashboard uses, are not for public use — they check for an internal secret.

02

It is a snapshot

Registry, GitHub and homepage data are refreshed on a schedule, so counts move during the day. The feed itself updates every five minutes.

03

Cite, don't launder

Stories are rewritten in our own words and link back to the publisher. If you republish, keep the original source address with it.