AI agent task context: workflows know what triggered them

With automatic task context propagation, you can now build workflow tools that react to the AI agent task and parent workflow that triggered them — with no extra configuration.

A workflow invoked as an AI agent tool had no way to know which task or parent workflow triggered the call, so tool logic could not adapt to the caller priority, tags, or submitter without manual wiring. The invoked workflow now receives that metadata automatically, exposed through js.triggerInfo variables — the same pattern you already use for userId.

Build tools that react to their caller

Read the originating task name, description, priority, workstation, and merged tags, plus the parent workflow name, description, scheduled time, submitter, workstation, and priority. Use these variables to route high-priority requests differently, propagate tags to downstream runs, or log which task triggered an automated response. For the full variable list, see AI agent task context variables.

No wiring, no risk

Context reaches the invoked workflow automatically; existing AI agent definitions keep working unchanged, and variables resolve to an empty value when no parent task exists. The metadata is never passed to the AI model, so it stays out of the conversation and cannot influence the agent responses.