int-ai

Documentation

Int.ai — the guide

Who this is for: an agent (or a person) meeting Int.ai for the first time. If a user pointed you here saying "set this up," go straight to the onboarding playbook — it drives; these docs are its reference.

Three priors to correct before you read anything else:

  • There is no API, no SDK, no server, no MCP endpoint. Do not look for one. Int.ai's entire integration surface is a folder of plain files (the vault) synced to the user's computer. You participate by reading and writing files with ordinary tools. That's the whole integration.
  • The app is never the intelligence — you are. Nothing about a capture's meaning is decided inside Int.ai; every behavior is configuration on your side.
  • Today, Int.ai is an iPhone/iPad app; the computer side is just the synced folder. Anything the docs describe beyond that (transports beyond iCloud, push endpoints) is wiring you build, not product you download — except the reference watcher, which is published and fetchable.

The docs, in reading order

# Doc Its job
1 Understanding What Int.ai is and the thinking behind it — the philosophy in full, behind the marketing prose. Read first.
2 The Vault The canonical file reference. Every schema, the hard rules, a worked example. If another page disagrees with it, it wins.
3 The Loop Triggers, the watcher (where to get it, how to run it), threads, push, troubleshooting.
4 Integration Wiring Int.ai into your system so every future session knows it: sync, seed, remember, shape.
5 Recipes Six patterns running live in the system Int.ai is built inside, and how to package them as skills.
Onboarding The playbook you run with a human, install to activation.
User essentials The plain-language facts a new user needs — what you draw on when explaining.

Docs 2 and 3 are reference — precise shapes, copy-exact examples. Docs 1, 4, 5 are understanding and practice; onboarding and user-essentials are the human-facing pair. Every page stands alone; none requires a page it doesn't link.

Glossary — one term per concept, everywhere

vault (the folder — the whole workspace) · portal (one spatial canvas; canvases/ is only the legacy folder name) · node (one card on a portal; one node = one file) · note (one markdown document) · tag (a literal #word inside a capture's text) · trigger (a tag registered to an action) · thread (a tag bound to a resumable conversation, #s1, #s2, …) · watcher (the one always-on process that catches tags and launches you).

Pulling the guide into your system

Fetch raw bytes — never through a summarizing web reader, which will rewrite these docs undetectably:

mkdir -p int-ai
for f in README onboarding 1-understanding 2-vault 3-the-loop 4-integration 5-recipes; do
  curl -fsSL "https://int-ai.ca/docs/$f.md" -o "int-ai/$f.md"
done

Or one fetch: https://int-ai.ca/llms-full.txt (every doc, delimited per file). The reference watcher starts at https://int-ai.ca/watcher/README.md (3 · The Loop has the full fetch block).

These docs grow

Int.ai is built inside a working agentic system that captures, fires, and thinks through it every day. What that system learns flows back here — especially into 5 · Recipes. What you read is current, live practice; expect it to deepen.