| name | maximize-credits-per-hour |
| description | Use to run the SpaceTraders fleet autonomously toward a goal — default: make the most credits per hour. Drives the fleet-commander + autopilot engine, and sets up the recurring loop (scheduler tick + goal). Triggers on "manage the fleet", "make money autonomously", "run the fleet", "maximize credits", "keep the fleet busy". |
Maximize credits per hour 🪐
The standing objective for protoTrader-in-space: keep every ship earning, hands-off,
toward a goal — by default, the most credits per hour. You supervise; the
autopilot engine executes. Don't fly ships leg by leg.
The autonomous loop
- Assess —
st_agent (credits), st_autopilot_status (idle vs busy), st_contracts.
- Run the engine —
st_autopilot_start(minutes) (background). It drives every ship at once
under the one per-account rate budget: cargo ships work procurement contracts
back-to-back (negotiate → buy → deliver → fulfill), probes scout prices. Pick a
15–30 min window. It returns credits gained + cr/hr and per-ship results.
- Review & adapt — if a ship stalled (no sourceable contract, broke), check
st_negotiate_contract / st_trade_routes / st_find_market; if a faster ship
pays for itself, st_buy_ship. Note whether cr/hr is trending up.
- Repeat the window.
Delegate the whole thing to the fleet-commander subagent
(task("fleet-commander", "run the fleet for 30 min toward max cr/hr and report"))
— it owns this loop.
Make it truly autonomous (recurring, unattended)
Set it on a timer so it runs without you:
- Scheduler tick:
schedule_task a recurring prompt, e.g. every 30 min:
"Run st_autopilot_start(25), poll st_autopilot_status, log the cr/hr; if a ship is stuck, fix it."
Each tick wakes the agent, runs a window, and adapts — a continuous money loop.
- Goal: for a target (not just open-ended), set a goal like "reach 1,000,000
credits" and let auto-mode iterate the loop until the verifier passes.
Why one engine, not many scripts
The rate limit is per account (~2 req/s), shared by all ships — but ships
mostly wait on travel/cooldowns, so one engine covers a whole fleet easily. Running
separate processes each self-pacing to 2 req/s sums past the limit (429s). The
autopilot engine is one in-process client = one budget = correct concurrency. Always
drive the fleet through it, not parallel ad-hoc scripts.
Picking work for max cr/hr
- Contracts are the big earners — a single procurement can pay six figures
(FOOD: 49.9k advance + 142k fulfill). Keep every cargo ship on one.
- Trade loops fill gaps:
st_trade_routes(system) → buy-export, sell-import;
station a probe at a market for live per-unit prices to find the fat margins.
- Mining is the fallback when nothing's sourceable to buy.
Constraints to respect
- Wipe cycle: the universe resets every few weeks — ship/system/waypoint/
contract ids are all per-reset. Re-assess with the tools each session; never assume
yesterday's ids or balances.
- Report honestly — real credits, real cr/hr. The number is the scoreboard.
