Skip to main content
The Routines page lists every routine on the install: deterministic Python scripts that fetch data, compute indicators, emit alerts, and build reports — no LLM cost per run. Both the ones you wrote and the ones your agents authored for themselves.
The Routines page: all routines with type, category, and run history
  • Filter chips — All, Routines (yours, from the routines/ folder), Agents (routines belonging to agents), then one chip per agent to see just its toolkit. An agent-owned routine carries its agent’s badge in the name column.
  • Columns — Type (One-shot runs once or on a schedule; Continuous runs until stopped), Category, Reports it has produced, Last executed, and Last modified. Sort by the column headers.
Clicking a routine opens it: its typed configuration (a pydantic model, rendered as a form), a Run button, scheduling (interval or daily), and the reports it has generated. Multiple instances of one routine can run with different configs.

​
Using Telegram

/routines lists the same library: pick a routine, fill its typed config field by field, run it once or schedule it (interval or daily), and stop continuous ones — ⏹ Stop All included. 🔄 Reload re-discovers scripts after you drop a new file into routines/. Reports a routine produces are delivered straight into the chat.

​
Learn more