Home / Integration
Integration
For legacy systems
and new ones alike.
No rewrite of business logic — but not zero changes either. You add a "bridge" to your system; the framework provides the rest.
What you add: a "bridge"
- An API to read business data and to receive AI proposals
- A manifest (the system describes itself)
- Each employee links their own chat — with your system's own login
- Recipient rules decided by your system
- Events written in the same transaction as the change
- Approval rules · scheduled reports (counts only)
Three ways to attach
- Node.js: mount server-kit on your server and wire hooks to your tables and functions
- Equivalent APIs exist: serve /bridge/v1 as aliases of your routes
- Other languages: implement OpenAPI 3.1 and prove it with the conformance suite
Born to work with AI employees
- Build on server-kit and the bridge comes with it
- You write only the business, hooks, manifest and screens
- A sample business system shows the layout and style
bridge.manifest.yaml # self-description src/auth.mjs # authenticate · isSessionLive src/domain.mjs # business rules · tx + emitEvent src/hooks.mjs # employees · recipientsFor · approvals src/routes.mjs # /api/* · readToolPolicy manual/NN-*.md # one screen = one section test/conformance.test.mjs web/ bff/ # web-kit · gateway
Keep the code, put the bridge outside
- Logins can be verified from outside (OIDC / SAML)
- APIs can be called as the user
- Changes are announced after commit (webhook / CDC)
- Approval and write APIs exist
AAll four → every feature
BAPIs, but not as the user → notices and reports
CScreens only → add APIs, or build anew
Checklist
What to add to your system (13 items)
Keep a minimal approval and scheduled report even if your business does not use them (so the conformance suite runs with nothing skipped).
| # | What to add | Kind |
|---|---|---|
| 1 | An API to read business data and to receive AI proposals | Required |
| 2 | Manifest (no warnings with --strict) | Required |
| 3 | A screen to link one's own chat (with your own login) | Required |
| 4 | Recipient rules decided by your system (same check as the web) | Required |
| 5 | Events in the same transaction as the change | Required |
| 6 | Staff and roles (including active / inactive) | Required |
| 7 | Approval rules in a single function | Required |
| 8 | Web session liveness check | Required |
| 9 | A staff manual (one screen = one section) | Recommended |
| 10 | Personal-data screening against your directory | Recommended |
| 11 | Editing the AI employee's profile | Recommended |
| 12 | Scheduled reports (counts only) | Required |
| 13 | External connectors | Optional |
Seven steps
Seven steps, and when you are done
1Write and validate the manifest
2Implement hooks (or alias existing APIs)
3Events in the transaction
4Conformance 52 / 52
5Register in the AI-plane profile
6Screens and BFF in your web app
7Enable automatic tasks
# 1 manifest bridge-validate manifest bridge.manifest.yaml --strict exit 0 # 2 conformance (TEST environment) node --test conformance/http.test.mjs 52 passed · 0 skipped # 3 your own tests node --test test/*.test.mjs all pass # 5–6 product profile · outbound settings closed node profile/scripts/check.mjs cordis.patch.yml exit 0 # 8 business web tsc --noEmit no errors
🎯
Start with one process
Begin with one — billing approvals, overdue deadlines — run it next to the existing screens, then widen.
📐
Size it first
Read APIs and login are the foundations; how much to add varies widely between systems.
🧩
Work with an AI agent
The integration skill guides AI coding agents through assess → plan → implement → verify.