No study yet: this module has no plate position.
What it does for you: It keeps the timetable for the proposed nightly and weekly checks of public sources (design). Each Product gets one check a night in a 20-minute slot between 02:00 and 05:00, UK time by default, with a fuller check once a week; the portfolio gets a weekly audit, a weekly proposal round and a Monday bundle for you. Every check has a fixed name, such as discovery/software-factory/basic/2026-09-29, so a restart finds the same check instead of starting another. Checks that were missed while the Mac was off or asleep are folded into the next one, and each keeps a final status; a Product added later has nothing counted as missed from before it was added. Before any AI model is asked anything, the schedule checks that the account's paid overage is recorded as off, that the week's token allowance has room, and that the same request was never sent before with an unknown result. When a usage limit is hit, the check waits and later picks up where it stopped, as the same check, and never sends again a request whose effect is unknown. The same holds when a stuck runner is confirmed gone: the check still waits for the limit to reset, or for your newer sign-in record, and gets no extra time the same night. A usage limit or a refused sign-in belongs to the account, so it pauses every check that uses that account, not only the one that met it.
How it links: It records everything in the Command journal. Intelligence declares that it uses no AI model. Nothing uses it yet: the source collector, triage and the nightly runner that would call it are later steps of the design.
Honest limits:
- Nothing runs on a schedule. No LaunchAgent is written or loaded: the module only produces the text of one, and refuses its timings unless the Mac's time zone is given as the commission's.
- Every time, window, deadline and allowance here is a proposal for your decision, not a signed commission. The weekly token ceiling has no default until a two-week trial run measures real use.
- Token counts are the Factory's estimates. The allowance limits what the Factory starts; it reserves nothing with a provider, and your own sessions share the same subscription limits.
- It cannot see whether a runner has really stopped. A stuck runner keeps its check until someone confirms it has gone; time alone never passes the check on.
- Records of your sign-in and paid-overage setting are statements, not proof: anyone who can write the journal could change them. That a live usage-limit message matches the test's example, and that paid overage can be turned off, are not established.
Status: In progress: implemented as increment 1 of the proposed discovery and post-development design, with its tests passing on this Mac. An independent adversarial pass found 12 defects, all fixed, and the fixed code then passed independent review after three rounds of further fixes; its separate reference review is still to come. It is local development only, not a release, a Verdict or a customer benefit, and all 27 epics remain open.