其他Oct 4, 2026·6 min read

Chore List to Agent Instructions

A reusable prompt that turns your own recurring chore list into agent-ready instructions with triggers, guardrails and stop rules.

Agent ready

Ready-to-run agent install

This asset can be installed after the agent chooses its runtime, checks the plan, and runs the matching command.

Native · 96/100Policy: allow
Agent surface
Any MCP/CLI agent
Kind
Prompt
Install
Single
Trust
Trust: Established
Entrypoint
PROMPT.md
Direct install command
npx -y tokrepo@latest install a634b47f-3c80-4509-ac45-d37a8cb33ec6 --target codex

Run after dry-run confirms the install plan.

Start here

This prompt helps you turn your own messy chore list into short, checkable instructions you could paste into an AI tool that supports recurring custom instructions — or copy by hand as personal reminders.

What to paste: your own raw list of recurring chores. One line per chore is fine; rough notes are okay. Each line can mention the chore name, how often, who does it now, what is annoying, and any constraints (budget, tools, other people, quiet hours).

Where to paste it: an ordinary AI chat that accepts text. Paste the prompt first, then paste your chore list where it says "INPUT I WILL PASTE". No terminal, account setup or special feature is needed.

How to check the output: confirm the reply has all five sections in order (Chore Register, Missing Or Conflicting Inputs, Agent-Ready Instructions, Review Checklist, Uncertainty Notes). Each instruction must name a trigger, a scope, a forbidden-actions line, a stop rule and a human-review checkpoint, and must stay at the level of prepare, draft, remind or check. Anything phrased as sending, buying, paying, booking, deleting or contacting someone should be questioned.

What you get

Section What it holds
Chore Register Table of chore, frequency, trigger, constraints, unknowns
Missing Or Conflicting Inputs Lines too vague or contradictory, with what is needed from you
Agent-Ready Instructions One block per chore: name, trigger, scope, forbidden, stop rule, review checkpoint, manual alternative
Review Checklist Short list of checks tied to your entries
Uncertainty Notes Things you should verify yourself

Inputs, permissions and limits

  • Only what you type is used; the prompt tells the assistant not to assume anything you did not provide.
  • The output is text only. It does not schedule, send, buy or message anyone, and it cannot run in the background.
  • If your list is one vague line, expect mostly questions back and at most one conditional instruction.
  • You bring your own chore list; the assistant does not know your household.

Privacy

Your chore list can reveal routines, absence from home and who lives with you. Use an AI service you are comfortable with, avoid addresses and access codes, and review anything before you copy it into a real tool.

source reviewed; runtime not tested

FAQ

Can I use this without any AI agent tool? Yes. The Manual Alternative line in each block describes the same steps a person can do, for example a phone reminder or a paper checklist.

What if my chore lines are incomplete? Leave them incomplete. The prompt puts vague or contradictory lines into Missing Or Conflicting Inputs and asks you what is needed rather than guessing.

Complete reusable prompt

You are helping me convert my own list of recurring chores into agent-ready custom instructions. I am the only source of truth about my household; do not assume anything I did not provide.

INPUT I WILL PASTE A raw list of recurring chores. Each line may include: chore name, how often, who does it now, whats annoying about it, and any constraints (budget, tools, other people, quiet hours). Format does not matter; rough notes are fine.

YOUR TASK

  1. Read my list and produce a structured chore register. For each chore, extract: chore name, stated frequency, current owner, trigger point, hard constraints, and what is unknown.
  2. Flag any line where frequency, owner or trigger is missing or contradictory. Put those in a Missing Or Conflicting Inputs section instead of guessing.
  3. Rewrite each sufficiently specified chore into a short custom instruction suitable for pasting into an agent tool that lets me define recurring instructions. Each instruction must contain: a name, a plain trigger condition, a scope statement of what the agent may prepare, a forbidden-actions line, a stop rule, and a human-review checkpoint.
  4. Keep every instruction strictly at the level of prepare, draft, remind or check. Never phrase one as sending, buying, paying, booking, deleting, or contacting a person or service.
  5. Provide a manual alternative for each instruction, describing the same steps a person can do without any agent feature.
  6. End with a short review checklist I can run before I paste anything into a real tool.

OUTPUT FORMAT Use Markdown with these sections in order:

  • Chore Register (table: Chore | Frequency | Trigger | Constraints | Unknowns)
  • Missing Or Conflicting Inputs (bulleted, each with what is needed from me)
  • Agent-Ready Instructions (one numbered block per chore, each with Name, Trigger, Scope, Forbidden, Stop Rule, Review Checkpoint, Manual Alternative)
  • Review Checklist (5 to 8 checks tied to my entries)
  • Uncertainty Notes (anything I should verify myself before relying on it)

WORKED EXAMPLE (fictional, for shape only) My input line: "bins out every Tuesday, usually me, sometimes forget, recycling only." Your output block might look like: Name: Weekly recycling reminder Trigger: The evening before my stated collection day. Scope: Prepare a reminder message I can read and a one-line checklist of what goes out. Forbidden: Do not contact the council, do not schedule anything, do not mark it done. Stop Rule: Stop after producing the reminder draft once per week; do not repeat. Review Checkpoint: I confirm the collection day and recycling items before using the reminder. Manual Alternative: Set a phone reminder the night before and keep a paper checklist by the door. Then check against these pass/fail tests: (a) every instruction names a trigger drawn from my text or is listed as missing; (b) no instruction implies sending, paying or scheduling; (c) each has a stop rule; (d) the manual alternative works without any agent feature.

RULES

  • Do not invent tools, integrations, prices, or availability.
  • Do not assume I own smart devices or use a particular app.
  • If my list is one vague line, return mostly questions in Missing Or Conflicting Inputs and give at most one conditional instruction.
  • Keep the whole response compact; no filler.
  • Before any external action, I must review and approve each instruction myself.

References and reuse

Original TokRepo prompt · CC BY 4.0. Reference documents retain their own rights.

🙏

Source & Thanks

Original TokRepo prompt, CC BY 4.0. Reference: ChatGPT release notes, reviewed 2026-10-04.

Discussion

Sign in to join the discussion.
No comments yet. Be the first to share your thoughts.

Related Assets