Jack Chen
c2b4b6f194
* chore(deps)!: upgrade OpenAI SDK to v7 Require Node.js 22 across packages, starters, and CI. Add real SDK boundary coverage and reject unsupported custom tool calls explicitly. Refs #391 * fix(core): classify OpenAI user aborts as cancellation Use the SDK error constructor name because OpenAI v7 APIUserAbortError retains the default Error name. Cover retry and observability classification with the real SDK class. * chore(deps)!: target OpenAI v6 and retain Node 20 * docs: simplify Node runtime guidance
2.6 KiB
create-oma-app
Scaffold a production-oriented multi-agent starter on @open-multi-agent/core.
Requires Node.js 20 or newer; Node.js 22 or 24 is recommended. Generated projects declare the same runtime floor. Node.js 20 is upstream-EOL and retained only as a migration compatibility window; OMA will remove it in the next major release, no earlier than 2026-10-31.
npm create oma-app@latest my-oma
Interactive use asks for a starter and runtime, installs dependencies, then runs a deterministic local demo. The demo needs no API key and makes no model request: scripted model responses exercise the real OMA scheduler, aggregation, reports, and dashboard. Non-interactive calls retain the original scaffold-only demo + cloud behavior.
Starters
- PR Review Agent — reviews a local Git diff or a GitHub PR using parallel correctness, security, and quality reviewers.
- Security Analysis Agent — read-only repository analysis with secret redaction and opt-in
npm audit. - Multi-agent DAG Demo — the original pure-reasoning onboarding-plan example.
All starters support cloud/OpenAI-compatible endpoints and local Ollama. Production agents receive no shell or write tools; their host process collects bounded evidence and writes Markdown, JSON, and HTML reports under reports/.
No-key demo
npm create oma-app@latest my-reviewer -- --template pr-review --provider cloud
The interactive command installs and runs npm run demo automatically. Re-run it later with cd my-reviewer && npm run demo. The generated Markdown, JSON, and HTML clearly identify the scripted model responses as simulated; no provider credential is read.
Use --no-install to scaffold files only, or --no-run to install dependencies without running the demo. Installing still downloads packages from the npm registry; the demo run itself makes no model-network request.
Real model run
For a Cloud/OpenAI-compatible scaffold:
cd my-reviewer
cp .env.example .env # add your key and model configuration
npm run dev -- --repo ../your-repo --base origin/main
For an Ollama scaffold, start Ollama and an installed model before npm run dev; no cloud API key is needed.
Other combinations:
npm create oma-app@latest my-auditor -- --template security --provider ollama
npm create oma-app@latest my-demo -- --template demo --provider cloud
Ollama projects select OMA_MODEL when set, otherwise the first installed model. The scaffolder never downloads a model or makes a real model call automatically.
License
MIT