n8n-nodes-niyoos

An n8n community node that lets you use Niyoos-governed LLMs
as a chat model inside n8n's AI features — the AI Agent, Basic LLM Chain,
Question and Answer Chain, and any other node that accepts a Chat Model.

Niyoos is an OpenAI-compatible governance proxy that sits
in front of your LLM providers (Anthropic and others). Your traffic is
authenticated with a per-agent key and is metered, cost-tracked, rate-limited
and audited centrally — while the wire format stays standard OpenAI
chat/completions. This node drives that proxy with the proven ChatOpenAI
client pointed at your Niyoos base URL, so every call your workflow makes is
governed by Niyoos automatically.

Installation · Getting a Niyoos agent key · Usage · Model dropdown · Known issues · License

What this package provides

• Credential — Niyoos API
  • Agent Key — your sk-aiwos-… key (stored masked).
  • Base URL — defaults to https://api.niyoos.com/proxy/v1.
  • The connection test calls GET /integrations/v1/me and passes when the key
    resolves to an active agent.
• Node — Niyoos Chat Model (a LangChain sub-node with an AI Language Model
  output; connect it to a root node such as the AI Agent)
  • Model — a dropdown populated live from GET /proxy/v1/models.
  • Options: Sampling Temperature, Maximum Number of Tokens, Top P,
    Timeout, Maximum Retries.

Installation

Follow the n8n community nodes installation guide.

In the n8n UI (self-hosted):

1. Go to Settings → Community Nodes → Install.
2. Enter n8n-nodes-niyoos and confirm.
3. Add the Niyoos Chat Model node to a workflow.

Manually (npm):

npm install n8n-nodes-niyoos

Compatibility: requires Node.js >=20.15 and a recent n8n with the
LangChain/AI nodes. Tested against n8n-workflow 2.16.

Getting a Niyoos agent key

1. Sign in to your Niyoos dashboard at niyoos.com.
2. Open Agents and create an agent (or pick an existing one). Each agent is
   bound to exactly one provider credential (e.g. an Anthropic key).
3. Reveal or create the agent's API key — it looks like sk-aiwos-….
4. In n8n, create a Niyoos API credential and paste the key. Leave the base
   URL at its default unless you run a self-hosted or staging Niyoos deployment.

Keep the key secret; it authorizes billed LLM traffic. You can revoke and
rotate it from the dashboard at any time.

Usage

1. Add an AI Agent (or any node with a Chat Model input) to your workflow.
2. From its Chat Model input, add a Niyoos Chat Model sub-node.
3. Select your Niyoos API credential.
4. Pick a model from the dropdown, set any options, and run.

Model dropdown: one agent = one provider

The Model dropdown is populated from GET /proxy/v1/models, which returns
the catalog for the specific agent your key belongs to. In Niyoos, one
agent is bound to exactly one provider credential — so the dropdown is
inherently scoped to that agent's provider. If your key belongs to an
Anthropic-backed agent, you will see Claude models only; you will not see
another provider's models here. To use a different provider, create a separate
agent (and a separate Niyoos API credential) for it.

Known issues

• Adding the Sampling Temperature option breaks claude-opus-4-8 and
  claude-sonnet-5. Anthropic has deprecated the temperature parameter on
  these models, so any request that sends it is rejected upstream with an HTTP
  400 (temperature is deprecated for this model). By default this node sends
  no sampling parameters — the fields only take effect once you explicitly
  add them under Options — so leave Sampling Temperature unset for those
  models. Other models (for example claude-haiku-4-5) still accept it.
• Dynamic connection labels are not supported. n8n names credentials by
  what you type; unlike some other integration platforms, it cannot label the
  connection from the identity response. The connection test still validates the
  key against /integrations/v1/me.

Resources

• n8n community nodes documentation
• Niyoos

License

MIT