nc://requesty

Requesty: an LLM gateway with managed routing policies, as a Nodecode provider. A Nodecode rewrite of requestyai/hermes-provider-requesty by requestyai.

With it, /connect offers Requesty, /models lists its routes, and a turn on requesty/<policy> goes to https://router.requesty.ai/v1 over the chat lane, as the original sends it:

• a model is one of Requesty's managed routing policies (a curated route with failover): /requesty reads GET /v1/models/managed and puts the ones that take tool calls in /models, and a bundled list of the original's fallback policies stands in until it does
• reasoning rides as the top-level reasoning_effort Requesty documents, and only to a model the catalog marks reasoning; Nodecode's minimal goes out as min, xhigh as high, and an off selection as none
• the session id rides as requesty.trace_id, so one conversation groups in the Requesty dashboard

See NOTICE.

Get it

Tell the agent "add the requesty cell", or run nodecode add nc://requesty.

Set it up

Make a key at https://app.requesty.ai/api-keys and save it with /connect, or set REQUESTY_API_KEY. A saved key wins over the variable.

The section is on unless it says "enabled": false:

{
  "requesty": {
    // where Requesty is served; the chat lane appends /chat/completions
    "base_url": "https://router.requesty.ai/v1"
  }
}

Use it

/requesty            # read the managed routing policies and list them
/requesty claude     # only the routes whose id carries "claude"

Then pick a model in /models and talk to it as usual. Point base_url at https://router.eu.requesty.ai/v1 to keep every request on Requesty's EU endpoint.

MIT licensed.