Run and pause the market analysis, and put the autopilot into read-only mode when something looks wrong. Every change asks for confirmation and is written to the audit trail.
Each run reads prices, Rhea pools and Burrow markets from mainnet, re-ranks the opportunities and re-analyses the demo portfolios.
Disabled on the public demo. Run the project locally to use it (README, Quick start).
Normal: the autopilot proposes and checks actions (dry run). Read-only: no plan may run at all, not even as a dry run; analysis continues. The system switches to read-only by itself on a safety trigger and never leaves it by itself.
| Started | Trigger | Result | Records |
|---|---|---|---|
| 10 min ago | scheduled | ✓Success | 387 |
| 25 min ago | scheduled | ✓Success | 429 |
| 41 min ago | scheduled | ✓Success | 31,669 |
The policy engine is independent of the agents: no agent and no text input can change its limits (value per action and per day, price freshness, agreement between price sources, slippage).
Newest first. Each change is a numbered entry in the audit trail.
| When | Portfolio | Change | Reason | By |
|---|---|---|---|---|
| 2026-11-08 05:23:01 UTC | Demo portfolio $5000 Shadow mainnet | new → ✓Normal | initial state (first start of this scope) | system |
| 2026-11-08 05:21:19 UTC | Demo portfolio $500 Shadow mainnet | new → ✓Normal | initial state (first start of this scope) | system |