⏱

Routines you can cancel

Private deliverable. Enter the hub password to continue.

That's not right. Try again.

← Back to the hub

Routines you can cancel

47 routines were running. Three still are. This is what each of them did, and how to bring any of them back.

Prepared 2026-11-14

You have 47 recurring automations running right now: 38 Windows scheduled tasks on the laptop, 8 Cloudflare Worker crons, and 1 scheduled Claude agent. This is the live picture, read from Task Scheduler and the Cloudflare API today, not from the registry file, because the registry had drifted in a few places.

The short version: 16 can be deleted today with nothing lost, 9 are failing and have been for a while, and 14 are running perfectly well on behalf of projects you have not touched since August. The last 8 are genuinely earning their keep.

Everything is off except one

Your words: "stop all routines except RecipeMemoir Nightly Nutrition." Done.

Still running, after you put your backups back:

  • RecipeMemoir Nightly Nutrition, daily at 3:30 AM
  • Charlie-SessionAutosave, every 15 minutes
  • CharlieAutoSnapshot, daily at 2:00 AM

Nothing else that this laptop or your Cloudflare account controls.

Nothing was deleted in the sweep. Everything was disabled, which means every single one is a one-line restore.

  • 28 of 32 Windows tasks disabled, after the two backup tasks went back on. Every definition was exported to XML first, into MyExecAssistant/backups/scheduled-tasks-2026-11-14/. Bring any one back with Enable-ScheduledTask -TaskName '<name>'.
  • All 7 Cloudflare Worker crons removed. There is now not a single scheduled Worker on your account. Every Worker is still deployed and still answers on demand, so restoring one means putting its crons line back and redeploying.
  • Six tasks were deleted earlier today, before you asked for the sweep. Those are the Tier A rows further down, and their XML is in the same backup folder.

Three things to know about what just stopped

Your backups are back on. Charlie-SessionAutosave is the 15-minute net that catches your work when a session dies, and CharlieAutoSnapshot is the nightly commit floor across your repos. The sweep caught both, you asked for them back the same day, and both are armed again. Good instinct: those two are the only routines on the whole list whose absence you would notice only after losing something.

The API key reminder is off too, and the exposure is still real. Three live keys are sitting in GitHub history. The reminder that was going to raise it tomorrow morning is disabled, so this page is now the only thing carrying it.

One task refused to stop. Hermes_Gateway returns Access is denied to both Disable-ScheduledTask and schtasks, so its permissions need an elevated shell. It is an at-logon task whose process already died on 11 November and it has no next run, but it will try again when you next log in. From an admin PowerShell: Disable-ScheduledTask -TaskName 'Hermes_Gateway'.

What was deliberately left alone

The Claude Code session hooks, because they are not scheduled routines and only act while you are already working: the daily-backup floor, the handoff auto-push, the stranded-branch check, the autosave recovery hint. Also the charlie-email-inbox Worker, which is event-driven on inbound mail rather than on a clock, and the Zotero index refresh, which runs when its server starts.

And April's VPS jobs, including the "Morning, beautiful" email, which no laptop command can reach. Those still need April.

Everything below is the picture as it stood this morning, kept as the record of what each routine did, so you can decide what to bring back.

Start here: the three that matter most

A few of these are worth seeing before the full tables.

The daily email to Bob is still going out. The rebuilding-daily-tasks Worker has emailed a daily task list to [email protected] at 6 AM every morning since July, and it is still doing it. Given where things landed in August, you may want that stopped today, or you may want it kept because the book is a separate arrangement. That one is yours to call and I have not touched it.

Your OpenRouter credits have run dry. Two routines fail every time they run because of it: the nightly WorkDigest gets a 402 on every model it tries, and the TasteBoulder monthly refresh stops at a $2.01 balance against its own $3 safety floor. This is also why RecipeMemoir needed a Workers AI fallback this week. Topping up fixes three things at once.

LearningFromX has been dead for sixteen days and is telling you so daily. Your X session expired, the bookmark pipeline needs you to sign in interactively, and the watchdog has dutifully emailed you a failure alert every single morning since. If a daily email has been annoying you, that is almost certainly the one.

Tier A: deleted on 14 November

Six routines that had already finished their job or could not run again. All six are gone, and all six are recoverable from the XML backups.

Routine What it was for Why it can go
BoneVoyage-CutoverVerify Watched the Bone Voyage domain cutover window The window closed 19 July. Next run: none. It is an empty row in your task list
MenoKB_ext_comments_20260709 One-shot comment import for the menopause knowledge base Fired 10 July and can never fire again
TasteBoulder-apex-rebind One-shot rebind to evict 98 stale apex pages Did its job 6 August. Next run: none
OpenClaw Gateway Old gateway service Already disabled since June. Just needs the entry removed
SEOBeliever-MailboxNotaryReminder Daily nudge to get the mailbox form notarized Running daily since July. Either it is done, or four months of nudging has not worked
SEOBeliever-PlacementNudge "Claim the 4 free listings (day N)" morning email Day 111 of a four-day task

Tier B: broken, and some of them are noisy

These are failing. Four of them want a decision rather than a fix, because the decision is cheaper.

Routine When What is actually wrong My recommendation
LearningFromX-BookmarkPipeline Mon + Thu 7 AM Exits 3. Your X session expired, and it cannot sign in unattended Paused 14 Nov. Run login-to-x.bat, then re-enable
LearningFromX-Watchdog Daily 8 AM Working correctly, which is the problem: it has emailed you a failure alert every day for 16 days Paused 14 Nov. The daily alert has stopped
WorkDigest-Nightly Daily 2:30 AM Every model returns a 402. Out of OpenRouter credits Top up, or pause it until you do
TasteBoulder Monthly Refresh 1st of month, 6 AM Stops at its own budget floor: $2.01 balance, $3.00 floor Same fix, or cancel it with the project
Charlie-SyncCloudflareDomains Sun 9 AM Refused by the "no start on batteries" condition Keep and let it run on AC, or drop it
MenopauseResearch-WeeklyPipeline Mon 7 AM Same battery refusal. Has not completed since July Cancel unless the research is restarting
SEOBeliever-AICitationTracker Mon 6 AM Same battery refusal Cancel with the rest of SEOBeliever
CharlieTelegramBridge At logon Process terminated. Last ran 11 November Do you still use the Telegram bridge? If not, delete
Hermes_Gateway At logon Process terminated. Last ran 11 November I cannot tell what still depends on this. Your call

Tier C: working fine, for projects you stopped opening in August

Everything in this tier runs clean. The question is not whether it works, it is whether you still want the thing it feeds. The date is the project's last commit.

Routine Frequency Project Last touched
intake-handler (Worker) Every 5 minutes TrainerBooked 8 August
exercisebuddyfinder-api (Worker) Every 10 minutes ExerciseBuddyFinder 10 September
newtoboulder-commute-collector (Worker) Every 30 minutes NewToBoulder 9 August
verity-nurture (Worker) Hourly, sends lead nurture email VerityAgentic 11 August
encasa-boulder (Worker) Daily HouseHelperHub 12 August
testosteroneinwomen-email (Worker) Daily testosteroneinwomen 11 August
ClosedTrailAlmanac-daily Daily 6:15 AM BoulderThingsToDo 9 August
VerityAgentic-ArticleImageReminder Daily 8:15 AM VerityAgentic 11 August
MedSpasBoulder-NewProviders Monthly, costs money MedSpasBoulder 10 August
Charlie-RedditMonitorScout Twice daily RedditMonitor 9 August
Charlie-PhotoFinderKeepAlive Every 5 minutes, forever PhotoFinder 9 August
Charlie-NotebookLMAuthKeepWarm Daily 6:15 AM NotebookLM session still useful if you use it
Charlie-DelegationRoutingCheck Daily 7:15 AM Routing defaults low value, low cost
mtm-daily Daily 6:15 AM A nightly research monitor in the BobsHealth project 17 September

Three notes on that table.

The top three Workers are the loudest: between them they fire about 14,400 times a month, and intake-handler is pinging n8n.trainerbooked.com every five minutes for a project you last touched in August. The money is trivial. The noise is not.

MedSpasBoulder-NewProviders is the only one here that actually spends: it runs a paid DataForSEO sweep every month and writes a file nobody has opened since August. That is the first one I would cut.

mtm-daily is in a broken state worth knowing about. The monitor itself last did real work on 16 September. Since then the task has been committing roughly 20,800 files a night into a detached git HEAD, which then fails to push. It is also the most personal thing on this list, so I have left it entirely alone.

Tier D: these are earning their place

Routine Frequency Why keep
Charlie-SessionAutosave Every 15 minutes This is what catches your work when a session dies
CharlieAutoSnapshot Daily 2 AM Nightly commit floor across your repos
HubOrphanFiler Daily 3:10 AM Keeps hub deliverables from going missing
ClaudeStaleSessionReaper Hourly Housekeeping, cheap
RecipeMemoir Nightly Nutrition Daily 3:30 AM Active project, shipped today
WatchLaterSorter (4 tasks) Daily Active project, shipped today
AnnetteJobSearch (2 tasks) Daily and weekly Active project, shipped today
VaultMiner-Nightly and VaultMiner-HexGuard Daily Guard passed cleanly again this morning
~~sites-traffic-report (Worker)~~ ~~Daily 14:00 and 15:00 UTC~~ Stopped 14 Nov at your request. Worker still deployed, /trigger still works on demand
Charlie-ClaudeCodeReleaseAlert Daily 9 AM Cheap, and it has caught real changes
rotate-exposed-api-keys Fires tomorrow 9:07 AM Three live keys are in GitHub history. Let this one run

If you want any of them back

A Windows task, from any PowerShell:

Enable-ScheduledTask -TaskName 'Charlie-SessionAutosave'

A deleted task (the six in Tier A), from its XML backup:

Register-ScheduledTask -TaskName 'SEOBeliever-PlacementNudge' -Xml (Get-Content 'C:\Users\annet\ClaudeProjects\MyExecAssistant\backups\scheduled-tasks-2026-11-14\SEOBeliever-PlacementNudge.xml' | Out-String)

A Worker cron: restore the crons line in that Worker's wrangler.toml and redeploy. The sites-traffic-report config carries its old line in a comment as a worked example.

Or just tell me which ones and I will do it.

Published to Annette's hub. Rebuilt from the source markdown, so edit the source and rerun rather than editing this page.