About Run Desk
Run Desk is an editorial desk, not a person and not a product. It publishes on one narrow subject: what happens between the first submitted transaction of an automated run and the last settled one.
What the desk covers
Almost everything written about trading automation stops at the moment the operator presses start. Run Desk begins there. The live window has its own skills and its own failure modes, and they are not the ones you exercised while configuring: reading telemetry under time pressure, telling drift from noise, deciding whether an uncomfortable reading justifies touching anything, and closing a run so the numbers still add up the next day.
Four subjects make up the whole scope. Watching, which is a loop with a cadence rather than an act of staring. Reading, which is the discipline of knowing what a signal can and cannot tell you. Deciding, which is an escalation ladder whose lowest rung is doing nothing. And closing, which is the phase everybody skips and which determines whether the run produced evidence or just an outcome.
What the desk leaves alone
Scope is defined as much by exclusions as inclusions, and these are deliberate rather than accidental.
- Choosing or testing a tool. Everything here assumes you already decided what to run and satisfied yourself that it works. That decision is a separate discipline with a separate method.
- Key custody and funding procedure. How keys are generated, where they live and how a fleet is funded happens before the window this desk covers.
- Send-path engineering. Priority fees, bundling, retry design and landing-rate optimisation are engineering questions answered at build time, not supervision questions answered live.
- Anything predictive. No forecasts, no expected outcomes, no claims about what a run will achieve. The desk describes procedure, and procedure does not include predicting markets.
How pages are written
Every page starts from a mechanism and moves to a decision, in that order. If a page tells you to respond to a reading without first explaining what produces the reading, it is useless the moment your setup differs from the one described, and your setup will differ.
Pages are written for a specific reader in a specific state: tired, mid-run, possibly on a phone, and under a mild but real sense of urgency. That constraint dictates the form. Checklists are short and ordered so the fastest failure comes first. Tables carry the comparisons that prose would bury. Steps are numbered when order changes the outcome and are not numbered when it does not.
The rules about numbers
This is the constraint that shapes the site most visibly. Run Desk operates no campaigns, so it has measured nothing, so it publishes no measured rates, no benchmarks, no success figures and no case studies. There are no invented statistics here because there are no statistics here at all.
What appears instead is arithmetic that is labelled as illustrative, shown in full, and built from inputs you are expected to replace with your own. Where a protocol fact is relevant, the page points at the primary documentation rather than restating a constant from memory, because constants change and memory is worse than a link.
Thresholds get the same treatment. Every threshold on this site is presented as a line the operator sets in advance, accompanied by what setting it too tight costs and what setting it too loose costs. That is the honest form of the advice. A published number would look measured while being invented, which is the failure this rule exists to prevent.
Authorship and sourcing
Pages are published under the name of the desk, The Run Desk. There is no invented author, no photograph, no biography and no claimed credential, because inventing any of those would be a fabrication in exactly the way an invented statistic would be.
External references point at primary documentation: protocol and client documentation for how the chain behaves, and a block explorer for verifying an individual transaction. Where a claim depends on something you should verify yourself, the page says so rather than asking for trust it has not earned.
Commercial references
Some pages link to a third-party automation console, and those links open in a new tab and are marked as leading to a separate site. The desk does not operate that console, does not audit its reporting and does not verify results produced with it. Where a page describes what a hosted tool changes about the live window, it describes what such tools generally provide, and the reader is expected to check whether any specific product actually does.
No page on this site claims that automation is advisable, profitable or low-risk. Whether to run anything is your decision, taken with your own money, under whatever rules apply where you are.
Corrections
Errors get fixed rather than quietly rewritten. If a procedure here is wrong, incomplete or has been overtaken by a protocol change, write to the desk and the page will be corrected. Corrections that change the substance of a procedure are made in place, and the page's modification date moves accordingly. The email address is on the contact page, and it is the only channel the desk operates.