Aster for n8n

Generate text with Aster Work through one Generate Text action. This community package has no runtime dependencies and uses n8n's credential system. It is published on npm; n8n verification is pending.

Install

This package requires Node.js 24 or later for consumers and development (engines.node: >=24). For self-hosted n8n, the Node.js process running n8n must meet that requirement; when using Docker, check the runtime inside the image. The qualified host is n8n 2.41.6, which also declares Node.js >=24. Development checks ran on Node 24.21.0; the mocked workflow import/execution passed in the official n8n 2.41.6 image on Node 26.7.0. These checks do not qualify every n8n or Node.js version allowed by the minimum requirement.

Self-hosted n8n users can install n8n-nodes-aster in Settings → Community Nodes. n8n Cloud discovery requires n8n verification. For local evaluation, build and pack this repository, then install the tarball in your disposable n8n custom-node directory:

From the repository root:

cd n8n
npm ci --ignore-scripts
npm test
npm run lint
npm run typecheck
npm pack

See n8n private-node installation guidance.

Configure

Create an Aster API credential using your existing Aster API key and select it on the Aster node. Credential testing uses authenticated GET /v1/models discovery; it does not generate billable text. Keys stay in n8n credentials. The API host is fixed to https://api.asterwise.dev.

Import the minimal workflow, choose your credential and execute only when you intend to use Aster credit. The example contains no credential identifiers or keys.

Generate Text

Each incoming item sends one non-streaming text request using aster-work at POST /v1/chat/completions. Prompt supports expressions. An optional system instruction precedes the prompt. Each text field allows at most 100,000 characters. Maximum Output Tokens defaults to 1024 and accepts integers from 1–4096; reasoning tokens can consume part of that allowance. This bounds output, not price or input context. No tools, attachments, saved chat or automatic retries are included.

Output contains text, model, finishReason, usage and id, with n8n item pairing. length indicates potentially incomplete output and does not trigger a retry. Usage metadata is not a billing receipt. The timeout is five minutes; inspect Aster Usage before retrying an uncertain request. Prompts and generated text may be retained in your n8n execution history.

Safe errors preserve these distinctions:

| HTTP | Meaning | | --- | --- | | 400 | Request rejected | | 401 | Invalid or revoked API key | | 402 | Insufficient funds | | 403 | Access denied or billing review | | 429 | Daily or concurrency limit | | 503 | Temporary service failure |

Continue On Fail emits a safe error and status code when available. Raw transport errors, authorization headers and request text are not forwarded by this node.

Development and qualification

npm test builds and runs mocked unit tests. npm run lint uses n8n's strict upstream rules. npm run typecheck checks TypeScript. Runtime qualification passed with n8n 2.41.6 in a disposable instance with HTTP requests intercepted and synthetic credentials; it is not live provider acceptance. See release preparation.

The development lockfile still includes upstream advisory findings in n8n tooling; see dependency review. No runtime dependencies are bundled.

API references: Chat Completions and parameters.

License

MIT, Copyright 2026 VectorQ Lab. Maintainer: [email protected].