Funding a wallet fleet
Moving SOL into many accounts without losing track of it: fan-out order, per-wallet floors, reconciliation and the sweep back when the run is finished.
Open the procedureRunbooks: procedures for the parts you cannot undo
A runbook is not a tutorial. It assumes you already decided to do the thing and now need it to go the same way every time, including the time you are tired and the time something has already gone wrong. Each procedure here states its preconditions first, names the step that cannot be reversed, and ends with a verification you can actually perform.
The four procedures in this section cover one full cycle: put value into the wallets, understand what the accounts themselves lock up, prove the configuration before it costs anything, and take the value back out without leaving accounts half open. Custody decisions that sit underneath all of it live in key hygiene.
Format Preconditions, numbered steps, verification, stop condition
Constants Protocol values with their formula, never from memory
Arithmetic Illustrative and shown in full, never presented as measured
4 procedures in this section
Numbered procedures with preconditions, rollback points and stop conditions. Funding a fleet, budgeting account rent, clearing a preflight and shutting a run down without stranding value.
Rent and account costs
What rent exemption is, why every token account locks lamports, how much a fleet actually ties up, and which of those lamports come back when you close accounts.
Open the procedurePreflight before a live run
The checklist between a configuration you believe in and a run you are willing to sign for: dry runs, simulation, one-wallet canary and the abort criteria.
Open the procedureShutting a run down cleanly
Stopping automation without orphaned transactions, stranded dust or half-closed accounts. Drain order, account closure, reconciliation and the archive you keep.
Open the procedureHow this section is used
Read them in loop order the first time: funding, then rent, then preflight, then shutdown. After that, treat each page as a reference you open at the moment you need the step, not something you memorise. If a procedure here disagrees with what your software actually does, the software wins and the desk wants to hear about it.
Key hygiene How keys are generated, where they live, what each one is allowed to touch, and how you rotate them without losing access to the accounts they control. Monitoring Signals worth alerting on, log lines worth keeping, and the ordered response for the moment a key or an endpoint is no longer trustworthy.