agentleFS
ARCASSystemsARCASSystems/FounderOS/skills/routines/SKILL.md

routines

Skill7 starsChanged 20 days agoMITView on GitHub

Discover, trigger, and understand your routines in plain English. Trigger on "what are my routines", "what runs automatically", "run my morning brief", "run my weekly review", "what should I change this month", "turn on the weekly review", or "how does the heartbeat work". Founder OS checks in when you open it (the SessionStart brief) and runs any routine on demand when you ask, all local with no account. Truly unattended while-you-sleep runs are an opt-in upgrade for users who have synced their OS to a remote, and the skill never claims a routine ran where it did not.

6 other projects use a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

4 rules. 0 are common, 3 are distinctive, and 1 is very specific to this project.

  • 0

    Common

    Most of the 6 edited versions we read keep these.

  • 3

    Distinctive

    This author's own choices. Worth a look.

  • 1

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    If core/identity.md does not exist, stop with: Founder OS not set up here.

    line 19
  2. Distinctive:

    Every routine runnable on demand with one plain-English ask.

    line 45
  3. Distinctive:

    Never claim a routine ran where it did not.

    line 66
  4. Distinctive:

    No em dashes or en dashes in anything you write.

    line 67

“Common” is measured: kept by at least 6 in 10 of the 6 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Routines

Runs on: reasoning - reads what routines exist and routes you to the one you asked for; the routines themselves run on their own mechanisms (below).

The heartbeat layer. Founder OS does not sit unused: it checks in when you open it and runs any routine the moment you ask. This skill is how you find, trigger, and understand them in plain English.

Pre-flight

  • If core/identity.md does not exist, stop with: Founder OS not set up here. Run /founder-os:setup first.

The heartbeat, honestly (hybrid)

There are two levels, and the difference is the whole point:

  • Free local floor (default, no account). Routines fire when you OPEN the OS (the SessionStart brief already does this) and ON DEMAND when you ask or run /loop. This is a real heartbeat for anyone who opens the OS even occasionally. It is fully local and truthful.
  • Unattended upgrade (opt-in). If you have synced your OS to a remote (start with the backup skill, then a remote routine on claude.ai), genuine while-you-sleep runs operate on the synced copy. This is the next level, never required, and only works because the remote can reach the synced files.

We never promise unattended local cron, and never claim a routine ran where it did not.

The routines

RoutineWhat it doesHow it runs
Daily briefFlags, stale cadence, decisions, quarantine, decay, connectors not set upOn-open (the SessionStart brief, automatic) + on-demand: say "run my brief" or /founder-os:today
Weekly reviewRolls the sprint, forces a verdict on every open flagOn-demand or on-open when a week has passed: say "run my weekly review"
Monthly flagshipThe three changes worth making now, each with a dated sourceOn-demand: say "what should I change" (the what-to-change skill). Freshness nag in the brief when a month has passed since the last run

To run any routine now, just say its name in plain English, or use the slash command in the table. To run one repeatedly in a session, wrap it with /loop (for example /loop weekly run my weekly review).

Turning routines on

A fresh install does NOT have unattended routines running on its own - that would be a false claim. What it has:

  • The on-open daily brief, active out of the box (it fires every time you open the OS).
  • Every routine runnable on demand with one plain-English ask.
  • An opt-in path to unattended runs once you sync (the upgrade above).

So the honest statement is: you can turn on any routine with one ask, and you can add unattended runs once your OS is synced.

Scheduling glossary (so the terms do not blur)

Four different things get called "scheduling". They are not the same:

  • Native CronCreate - in-session only. Fires while the Claude Code REPL is idle, and a recurring job auto-expires after 7 days. Good for a short in-session cadence, not a reliable weekly or monthly cron.
  • RemoteTrigger (claude.ai routines) - runs on claude.ai infrastructure. It cannot read your laptop, so it only helps once your OS is synced to a remote it can reach. This is the unattended upgrade.
  • Cowork scheduled tasks - from 6 October 2026 new Cowork tasks on Pro and Max plans run in Anthropic's cloud. Anthropic's help pages disagree on whether a scheduled task can use local files: one says it cannot be tied to a folder on your computer, another says it can while the desktop app is open. Treat a Cowork schedule as unreliable for your OS folder. Use it for jobs that need your connectors alone (an inbox summary, say), and keep Claude Code as the OS layer.
  • The old external task-runner MCP is no longer a dependency. Founder OS uses the native on-open / on-demand floor and the optional remote upgrade instead, so there is nothing extra to install.

Runtime honesty

On a web-only surface, the on-open brief and any local-script routine cannot run; say so and offer the on-demand version you can do in-session (reading files and reasoning). Never report that a routine ran on a surface that cannot run it.

Rules

  • State each routine's real mechanism. Never imply unattended local cron exists.
  • Never claim a routine ran where it did not.
  • No em dashes or en dashes in anything you write. Hyphens only.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.