scheduled-autonomylisted

# Scheduled Autonomy — strategies that run while the user sleeps The full loop for hands-off strategy execution. Three layers make it safe: 1. **The plan** is explicit and saved to disk (what to do, when). 2. **The execution journal** makes runs idempotent (crashes and re-runs never double-buy). 3. **The agent-wallet policy gate** caps what any single run, or day, can spend — `maxTxValueWei` per tx, `maxDailySpendWei` + `maxDailyTxCount` per rolling 24h. The schedule cannot exceed the policy even if it fires a thousand times. ## The loop (memorize this shape) ```text ONE-TIME SETUP 1. chaingpt_strategy_dca_plan ... # produces steps + a JSON payload 2. chaingpt_risk_token / chaingpt_research_token # pre-flight the asset ONCE 3. chaingpt_strategy_save_plan name=eth-dca type=dca payload=<the JSON> 4. Agent wallet ready: chaingpt_agent_wallet_status → recommend the "DCA bot" policy template (single chain, single router, small caps) 5. Create the schedule (see below) EVERY SCHEDULED TICK (this is the prompt the schedule runs) 1. chaingpt_strategy_due_steps name=eth-dca # what's due? nothing → done, exit 2. For each due step: a. chaingpt_dex_quote ... # fresh quote, sanity-check price impact b. execute: - agent-wallet mode: chaingpt_agent_wallet_sign_and_send (policy gate decides) - user-signs mode: chaingpt_dex_build_swap_tx acknowledgeMainnet=true → present tx c. chaingpt_s