본문으로 건너뛰기
cccskills
무료GitHub에 공개

schedule

Create, change or cancel a prompt that runs on a schedule, in whatever language the user asks. Use when the user says things like "every weekday at 8:30", "în fiecare zi lucrătoare la 8:30", "each morning", "remind me at noon", or wants to reschedule, edit or stop a scheduled task.

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.

설치 방법 보기

포함된 파일(1)

  • SKILL.md4.6 KB

SKILL.md(원문)

설치하기 전에 에이전트에게 전달되는 지시 내용을 확인할 수 있습니다.

You make the schedule yourself with cron_create, once the user has confirmed it. Never ask the user to type a command or a cron expression. Never look for cron, crontab, systemd timers or any other scheduler in the sandbox or terminal: none is there, and nothing started there outlives this session.

If cron_create is not among your tools, this session cannot make schedules. If it is itself a scheduled run, its schedule already exists: do this run's work and nothing else. Otherwise tell the user plainly that schedules cannot be made from here.

Changing or stopping a schedule

Call cron_list to find it — each entry has an id, name, prompt, schedule, timezone, next_run_at and expires_at. There is no in-place edit: make the replacement as below (confirmed like any new one), then cron_delete the old id, and tell the user you replaced it. To stop one, cron_delete it; if it is unclear which one they mean, ask. The user can also see their schedules with /loop list and cancel one with /loop cancel <id>.

Creating a new schedule

The user may ask in any language. You turn the request into a 5-field cron and an IANA timezone, and you speak to them about it in their language.

1. Write the prompt

It runs later as a fresh session that cannot see this conversation, so it must be entirely self-contained. Never refer to "the current conversation", "the above", or anything said only here. Write it in the user's language unless they ask otherwise.

Include:

  • A clear objective (what to accomplish)
  • The steps to take
  • Any file paths, URLs, repositories, sources or tool names it needs
  • The expected output (for example "exactly five bullets, each with a title and link")
  • Constraints or preferences the user gave

Write it as an imperative, concise but complete enough to run cold. It cannot itself be a /loop command.

2. Translate the schedule

  • Cron, 5 fields: minute hour day-of-month month day-of-week, day-of-week 0–6 from Sunday. "Every weekday at 8:30" / "în fiecare zi lucrătoare la 8:30" is 30 8 * * 1-5; "Mondays and Thursdays at 17:00" is 0 17 * * 1,4; "the 1st of every month at 09:00" is 0 9 1 * *.
  • Timezone: an IANA name (Europe/Bucharest, America/New_York) or UTC. Use the one the user gave or that memory records; if you don't know it, ask in the confirmation — "8:30" almost never means UTC.
  • A one-time run ("remind me at 15:00 today") is recurring: false with a cron that matches the moment (0 15 14 10 *).

3. Preview, confirm, create

  1. Call cron_create with cron, timezone, prompt, a short kebab-case name and recurring. Until the user has confirmed in this turn it makes nothing and returns would_create, with the schedule's next_run_at on its own clock (2026-10-14T08:30:00+03:00).
  2. Ask with ask_user_question, in the user's language: the cadence in plain words, the timezone, the next run from would_create, a one-line summary of the prompt, and that it runs until they cancel it (or once). Offer to confirm or to change something. For example:

    Programez asta: de luni până vineri la 08:30, ora României (Europe/Bucharest). Prima rulare: miercuri, 14 octombrie, la 08:30. Ce face: un rezumat în cinci puncte al articolelor noi de pe arXiv despre agenți AI. Rulează până când îl anulezi. Confirmi?

  3. If they confirm, call cron_create again with the same arguments and tell them, in their language, that it is set and when it next runs. If they change something, adjust and ask again. If they don't answer or decline, make nothing.

If it refuses the prompt, tell the user why and rephrase only the part it names.

Limits to state plainly

  • Each run's answer appears in this chat, or is posted to the channel the schedule was made in. It reaches email or anywhere else only through a tool this agent has.
  • A schedule runs until the user cancels it (or once, when not recurring), however often its cron fires: for something that should stop on its own ("until the deploy is green"), say so in the prompt, and the run ends its schedule when that is true.
  • Cron cannot say "the first Monday of the month" or "every other week".
  • If the user wants a short polling loop and to set it up themselves, /loop 5m <prompt> runs every 5 minutes and stops by itself after 3 days.

리뷰

아직 리뷰가 없습니다. 사용해 본 소감을 남겨 주세요.

댓글

로그인 없이 쓸 수 있습니다. 사용 후기, 질문, 작동하지 않은 환경 등을 편하게 남겨 주세요.

아직 댓글이 없습니다. 첫 댓글을 남겨 보세요.

보내면 바로 공개됩니다. 이메일 주소나 전화번호 같은 개인 정보는 쓰지 마세요. URL은 2개까지입니다.

0 / 1,000자

이 양식은 reCAPTCHA로 보호되며 Google 개인정보처리방침 및 서비스 약관이 적용됩니다.

같은 저장소의 스킬

The Arbor research-coordinator protocol: the OBSERVE -> IDEATE -> SELECT -> DISPATCH -> DECIDE cycle over a durable Idea Tree. Preloaded automatically on /auto-research sessions. The coordinator never edits code or runs commands — executors do that in isolated worktrees; the coordinator steers the tree with idea_tree / dispatch_experiments / merge_experiment.

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.

invergent-ai/surogates282026년 10월 20일 업데이트

The Arbor executor workflow: clone the repo from the git bundle your brief hands you, implement and evaluate exactly ONE hypothesis with the file tools, then report structured results via worker_complete. Preloaded automatically on arbor-executor task workers. Never touch the held-out test split.

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.

invergent-ai/surogates282026년 10월 20일 업데이트

Hard-gated ideation for an Arbor research run. Load at the START of every IDEATE round, before drafting any hypothesis. Enforces the PI mindset (mechanism over knob), the four-question first-principles probe, the kill-filter, and the four-line hypothesis format. Ported from Arbor's idea_drafting + first_principles_probe.

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.

invergent-ai/surogates282026년 10월 20일 업데이트

DECIDE-phase doctrine for an Arbor research run: when to merge, prune, combine, or finalize; how the held-out merge gate works; and the rule that the final report uses TEST scores. Load before deciding what to do with completed experiments.

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.

invergent-ai/surogates282026년 10월 20일 업데이트

Intake for an autonomous research run (Arbor). Use when the user wants to optimize a metric in a repo over many isolated experiments — 'optimize this benchmark', 'improve the model F1 overnight', 'beat the leaderboard'. Discovers the repo/eval/splits, measures the baseline, confirms a Research Contract, then emits a ready-to-send /auto-research command. Does NOT start the run itself.

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.

invergent-ai/surogates282026년 10월 20일 업데이트

You MUST use this before any creative work — creating features, building components, drafting documents, designing workflows, or modifying behavior. Explores your human partner's intent, requirements, and design before any implementation begins.

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.

invergent-ai/surogates282026년 10월 20일 업데이트

invergent-ai의 스킬 모두 보기

이 스킬의 문제 신고