agentleFS
saypage-aisaypage-ai/saypage-plugin/skills/dm-writing/SKILL.md

dm-writing

Skill0 starsChanged 12 days agoMITView on GitHub

Write the messages and button labels of a Saypage comment-to-DM automation (opening message, link message, follow request, email request, public reply), and help answer someone who wrote first. Saypage sends DMs only to people who comment a keyword and cannot read the inbox. Use for requests like "write the messages my automation sends", "what should the DM say", "ask for their email in the DM", "how do I answer this person".

The file

We split this skill into the rules it gives an agent

1 rule. About 0 are common, 0 are distinctive, and 1 is very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈0

    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. Very specific:

    Run every message through natural-voice before showing it.

    line 81

≈ “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

DM writing

A DM lands in a personal inbox, next to messages from friends. It is welcome when the person asked for it (they commented the keyword, replied to a story, asked a question) and it delivers what they asked for, straight away. It is unwelcome in almost every other case.

What Saypage can do here

Saypage sends DMs only through keyword automations: someone comments the keyword under a post and gets the automation's messages. It cannot read or summarise the inbox, send a one-off message, or write to people who did not comment. When the user asks to read their DMs or to message followers in bulk, say so and call no Saypage tool for that: offer to help with messages they paste, or a keyword automation instead.

1. Automation messages (Saypage)

Instagram requires two steps: an opening message with a button, then the link.

  • opening_message: thank them for the comment, say what is coming, ask for the tap. "Here's the croissant recipe you asked for 🥐 Tap below and I'll send you the link."
  • opening_button_label: what the tap gets them. "Send me the recipe".
  • link_message: deliver at once, a tip for using it, and optionally one easy question. "Here it is! The dough needs a night in the fridge, so start it tonight. Which filling are you trying first?"
  • link_button_label: "Open the recipe".
  • extra_link_buttons (up to 2 more): the shop, the booking page, the YouTube channel, only when the follower would want them.
  • With follow_gate: true, follow_request_message and follow_request_button_label: say plainly that the link is for followers, without guilt. "This one's for my followers: follow @maisonlune.lyon, then tap below and it's yours." / "I'm following". A follow gate wins follows and loses some of the people who wanted the link: say so to the user.
  • With email_gate: true, email_request_message: say what they get and what the address is for. "Where should I send the guide? Reply with your email and you'll get it, plus my recipe of the week." It comes after the tap and any follow gate, and they type the address in reply. The link goes out once a reply holds one; the request goes out 3 times at most, then the sequence stops. Followers who already gave this account an address get the link at once. An email gate builds the list and loses some of the people who wanted the link: say so to the user.
  • public_reply (optional): the short public answer under the comment ("Sent you a DM!").

Rules for all of them:

  • Button labels: 20 characters at most. Count them.
  • Every message in the user's language, including the follow-gate and email ones when those gates are on: the defaults are English, and a DM that switches language looks broken.
  • No surprise conditions. The link comes with the second message unless the user chose a follow or email gate. With a gate, say so in the ask ("Comment GUIDE and leave your email") and say what they get for it: a condition that only shows up in the DMs makes a free offer feel like a trap. Ask for nothing else in return, such as a story or a share.
  • Let the user choose the words. These DMs speak in their name: offer two tones (warm, short) or let them write their own, and show the final texts before create_automation or update_automation, as the saypage skill describes.

To see where people stop and rewrite the step that loses them, use funnel-review.

2. Messages the user sends by hand

Saypage only helps write these; the user sends them from the app. When someone wrote first (a story reply, a question), answer what they asked in two to four sentences, then one small question or next step. Keep every fact true, and never automate outreach: many cold messages in a row get accounts restricted.

Output

Each message ready to copy, with its character count, labels with their length:

opening_message       "Here's the croissant recipe you asked for 🥐 Tap below..."   (81)
opening_button_label  "Send me the recipe"                                           (18/20)
link_message          "Here it is! The dough needs a night in the fridge..."         (107)
link_button_label     "Open the recipe"                                              (15/20)

Run every message through natural-voice before showing 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.