SentinelProxy icon

SentinelProxy

Send an AI agent turn (system prompt, user input, response) to SentinelProxy for self-healing evaluation

SentinelProxy

AI-generated

Summary

Sends each input item’s agent turn to SentinelProxy for evaluation by POSTing the system prompt, user message, model response, and latency to /api/v1/ingest. Requires SentinelProxy API credentials; the node returns the endpoint response for each item.

Inputs

  • System Prompt — The agent’s system prompt.
  • User Input (required) — The user message for this turn.
  • LLM Response — The model output for this turn; recommended but optional.
  • Latency — LLM call round-trip latency in milliseconds; defaults to 0.
  • SentinelProxy API credentials (required) — Credentials used to authenticate the request. The base URL defaults to https://sentinelproxy.tomphillips.uk if not set.

Output shape

One output item per input item, containing the API response in its json field.

Each item is paired with its corresponding input. If Continue On Fail is enabled, a failed request returns { "error": "..." } for that item; otherwise the node raises an error.

Links

Discussion