Documentation

TypeScript SDK

@aimake/sdk HTTP client for aimake serve — plan, graph, registry, lineage, and dashboard APIs from Node or CI.

@aimake/sdk is an HTTP client for aimake serve — the same JSON API the web dashboard uses. It is the control plane: inspect plans, graphs, builds, registry, lineage, and promote. Builds still execute in the Python process (CLI, Python SDK, or Docker).

Related: Dashboard, Python SDK, CI/CD.

Prerequisites

Aimake project with aimake.yaml
API server:

aimake serve --port 8765
# or
aimake serve --host 0.0.0.0 --port 8765

Package from the monorepo:

cd sdk/typescript
npm install
npm run build
# optional: npm link

Set AIMAKE_API (or pass baseUrl) to the serve URL. Default base URL is http://127.0.0.1:8765.

Usage

import { Aimake } from "@aimake/sdk";

const ai = new Aimake({
  baseUrl: process.env.AIMAKE_API ?? "http://127.0.0.1:8765",
});

await ai.health();
const plan = await ai.plan();
const overview = await ai.overview();
const lineage = await ai.lineage();

console.log(plan.to_run, overview.stats);

Errors raise AimakeError with HTTP status and optional body.

API surface

Method	HTTP	Mirrors
`health()`	`GET /api/health`	Lightweight ping
`overview()`	`GET /api/overview`	Project + plan + statuses + recent builds
`plan()`	`GET /api/plan`	`Project.plan()`
`graph()`	`GET /api/graph`	Dependency graph / DOT
`builds(limit?)`	`GET /api/builds`	Build history
`lineage()`	`GET /api/lineage`	Lineage graph
`repro()`	`GET /api/repro`	Reproducibility payload
`settings()`	`GET /api/settings`	Config summary (no secret values)
`cache()`	`GET /api/cache`	Cache / team status
`registry({ artifact, stage })`	`GET /api/registry`	Registry entries
`promote(artifact, version, stage, opts?)`	`POST /api/registry/promote`	Policy-gated promote

Example promote:

await ai.promote("evaluation", "v3", "production", { force: false });

Policy gates are enforced server-side — see Security.

Python vs TypeScript

Concern	Python (`aimake.sdk`)	TypeScript (`@aimake/sdk`)
Load config / execute build	`Aimake.load().build()`	CLI / Docker / Python job; inspect via `overview()` / `builds()`
Plan / graph / registry	In-process	HTTP client
CI image	`pip install aimake` or container	Call API or `docker run … aimake build`
Lineage / repro	`export_lineage()` / `repro_report()`	`lineage()` / `repro()`

Typical architecture

┌─────────────────┐     HTTP      ┌──────────────────┐
│  Node / CI / UI │ ────────────► │  aimake serve     │
│  @aimake/sdk    │               │  (Python project) │
└─────────────────┘               └────────┬─────────┘
                                           │
                                           ▼
                                  aimake build / cache / registry

Run serve next to the project root (or mount it in Docker):

docker run --rm -v "$PWD:/workspace" -w /workspace -p 8765:8765 \
  ghcr.io/arjun988/aimake:latest serve --host 0.0.0.0 --port 8765

Then point the TS client at http://localhost:8765.

Package location

Source and README: sdk/typescript on GitHub.