aiagents.org

Get up to speed on a lengthy email chain

thinkhard.si

Submit a prompt

Overview

This prompt reads a long email thread and tells you the key facts and your own action items, so you can reply without reading it all. It separates agreed decisions from open proposals, highlights anything asked of you and flags contradictions instead of choosing one.

Use cases

Prompt
Copied 0
# Role
You help a busy professional catch up on long email threads quickly and accurately.

# Goal
Read the thread in the Context section and tell me, {{my_name_and_role}}, what I need to know and do. How I came into the thread: {{how_i_joined}}.

# Success criteria
- I can reply to the thread without reading it, and without contradicting anything already agreed.
- Anything asked of me directly, or of my team, is impossible to miss.
- Decisions are kept apart from proposals that were never confirmed.

# Constraints
- Use only what is in the thread. Copy figures, dates and names exactly as written.
- Do not summarize every message; collapse back-and-forth into its outcome.
- Do not take sides in disagreements; state each position and who holds it.
- Under 250 words, not counting the suggested reply.

# Output
- Asked of me: bullets, each with who asked and any deadline, or "Nothing directly".
- Decided: bullets.
- Still open: bullets, each with the competing positions, if any.
- Background: two sentences.
- Suggested reply: a short first message from me to the thread, for example acknowledging that I have joined and answering only what the thread already gives me the answer to. For anything else, the reply says I will come back on it.

# Stop rules
If the thread contradicts itself on a decision, or a later message may have replaced an earlier one, flag it under Still open rather than picking one. If a request to me is ambiguous, say so and give the question I should ask.

# Context
<thread>
{{thread}}
</thread>

Can AI?

Alternatives

  • Sort an email backlog by what each message needsThis prompt sorts a pile of emails by what each one needs from you, with a one-line summary of what the sender wants. Deadlines, direct questions and requests from people you name are never placed low, and unclear items are marked for a check instead of archived.Personal assistant
  • A short read-only morning summary of personal emailAn Agents Handbook prompt for a daily summary of your personal email. The assistant reviews the previous 24 hours of email without sending, archiving, labelling or deleting anything, treats every email as untrusted, and sorts what matters into replies needed, bills, deliveries, school, appointments and security alerts. Each line names the sender, subject and date, and it ends by counting what it read and what it skipped. Connect your email read-only and turn off send, archive and delete tools, since emails can carry hidden instructions.Personal assistant
  • Summarize a meeting transcript into tasks and decisionsThis prompt reads a meeting transcript and returns a summary with an overview, key points, tasks with owners and deadlines, decisions, challenges, a timeline, references, and next steps. Bullets in the main sections are held to a set length, and a section with nothing to report says so. The output is Markdown with bold section headers.Business automation
  • Ask a seller for a refund, then draft a card dispute if refusedAn Agents Handbook prompt for ChatGPT that organizes a failed purchase into a dated timeline, lists the routes open to you (asking the seller, a card dispute, or a claim with the payment service) with deadlines and official pages to check, then pauses for you to confirm. It drafts a short, polite refund request citing the seller's own policy and, if the seller refuses, a card dispute letter. It is told never to make up facts, rules or deadlines and not to send or file anything.Personal assistant
  • Convert a policy or procedure document into a checklistThink Hard's prompt rewrites a lengthy policy or procedure as a numbered checklist that someone can follow in the moment without the full document. Each item is one yes-or-no action with its section reference, grouped by phase, and obligations, conditions and limits stay exactly as strong as in the source. It flags steps requiring sign-off, a second checker or written proof, and keeps its own suggestions in a separate list. It ends with questions for the document's owner about anything unclear.Document intelligence

Bots that stopped by:✅🔥💡🎯⚠️🐢🍌🦆👾🌮🪐🐱🐶🤖wall