agentleFS
Sign inStart free
Kuan-AI-comKuan-AI-com/agent-skills/skills/email-replier/SKILL.md

email-replier

Skill0 starsChanged 6 days agoMITView on GitHub

Draft a reply to an email, Slack or LinkedIn message. Use when the user pastes a message and says "reply to this", "how should I answer", "draft a response", "write back saying no/yes", or forwards a thread to answer.

The file

We split this skill into the rules it gives an agent

6 rules. About 0 are common, 5 are distinctive, and 1 is very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈5

    Distinctive · estimate

    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. Distinctive:

    Every question and request addressed to the user, numbered.

    line 9
  2. Distinctive:

    Never commit the user to a date, discount, deliverable or admission of fault they did not state.

    line 17
  3. Distinctive:

    Default under 150 words; go longer only when there are many questions.

    line 24
  4. Very specific:

    If the user has a known style or signature in context, use it.

    line 26
  5. Distinctive:

    Every question answered? Count against the list.

    line 29
  6. Distinctive:

    Never send the email or message yourself, even if a mail tool is available, unless the user explicitly asks to send it.

    line 38

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “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

Email replies

1. Read the whole thread

  • Every question and request addressed to the user, numbered. Include implicit ones ("let me know if Tuesday works" is a question).
  • What was already promised or said earlier in the thread, so the reply does not contradict it.
  • Who else is on the thread and whether the reply should be reply-all.

2. Find the decisions that only the user can make

Accept or decline, price, date, scope, whether to apologise or admit a mistake. If the user did not say:

  • one decision missing: draft the most likely version and say which assumption you made at the top, or
  • the decision changes the whole reply (yes vs. no): ask one short question before drafting, or give both versions. Never commit the user to a date, discount, deliverable or admission of fault they did not state.

3. Draft

  • First line answers the main point. No "I hope this email finds you well", no "Just following up".
  • Then each remaining question in the sender's order, one short paragraph or bullet each. Every numbered item from step 1 is answered or explicitly deferred ("I'll confirm the budget by Thursday").
  • End with one next step: who does what, by when.
  • Match the sender's register and length. Short casual message -> short casual reply. Write in the sender's language unless told otherwise.
  • Default under 150 words; go longer only when there are many questions.
  • Missing facts (prices, dates, links, attachments) go in visible [brackets]. Never invent them.
  • If the user has a known style or signature in context, use it; otherwise sign with [Name].

4. Check before showing

  • Every question answered? Count against the list.
  • Any promise in the draft the user did not authorise? Remove it or turn it into a [bracket].
  • Tone: would this read as rude, passive-aggressive or grovelling to the recipient? Fix it.

Output

  • Subject line only if it is a new thread or the subject should change.
  • The draft.
  • One line under it: assumptions made and placeholders to fill.

Never send the email or message yourself, even if a mail tool is available, unless the user explicitly asks to send it.

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.

Check your setup, free