AI With Amirthan

← All guides

Content · July 11, 2026 · 2 min read

Triage your Instagram DMs with AI that never replies

Let an AI agent read and sort your Instagram DMs, flag who needs a reply, and never send a word. The read-only triage pattern that keeps your replies human.

The problem

Your Instagram DMs are where a follower becomes a community member or a customer, and they are the easiest surface to fall behind on. Real questions sit buried under reel shares and casual threads. The tempting fix is to hand the inbox to an AI agent and let it answer. That is the one automation I refuse to build: an agent replying in your voice is one bad output away from telling a real person something you never said.

What I built instead

I let an agent read and triage my Instagram DMs while sending stays strictly human. The agent connects through Composio, an integration hub with a free tier, using a one-time CLI login, so no API key sits in any config file. The skill is limited to exactly three calls: get account info, list conversations, list messages. In its world, send, delete, and mark as seen do not exist.

When I ask which DMs need a reply, it scans my most recent threads and returns a digest in three buckets: needs a reply (they messaged last), waiting on them (I messaged last), and everything else. It flags anything time sensitive, like a question someone asked or a resource I promised to send. Then I open Instagram and answer the short list myself.

An agent that can read your inbox but cannot speak in it can never embarrass you.

Why not let the AI reply to DMs?

Because a DM is the moment someone checks whether you are real. Two rules keep the pattern safe:

  1. Replies stay human. The agent may draft a reply for me to review, but sending is always my finger on the button.
  2. Triage is pull only. I run it when I ask for it, never on a schedule. My competitor research runs on autopilot because it only reads public pages. An inbox is private, so it never goes in a loop.

Copy the DM triage pattern

Connect your Instagram account to your agent through an integration hub with read access, then paste this:

Read my Instagram DM inbox through the connected read-only
integration. You may list conversations and read messages.
You may not send, delete, or mark anything as seen.

Return a triage digest, newest first, in three buckets:
1. Needs a reply: threads where the other person messaged last.
   Flag anything time sensitive, like a question or a promised link.
2. Waiting on them: threads where I messaged last.
3. Everything else: media shares and casual threads.

You may draft replies for me to review. Never send anything.

Do this now

  1. Connect Instagram to your agent through an integration hub, read access only.
  2. Paste the triage prompt above and read your first digest.
  3. Answer the needs-a-reply bucket yourself, today.

Claude Code

Turn this into a Claude skill

Paste this into Claude Code. It reads this guide and builds a skill shaped to how you work. 

Read the guide at https://aiwithamirthan.com/guides/triage-instagram-dms-without-letting-ai-reply. Then build me a Claude Code skill from it, personalized to how I actually work.

1. Fetch and read the full guide at the URL above.
2. Look at my setup so the skill fits me, not a generic template: skim my repos, my CLAUDE.md / AGENTS.md files, my existing skills, and my stack.
3. Create a new skill (a SKILL.md with a clear name, description, and steps) that lets me do what this guide teaches, adapted to my tools and goals.
4. Tell me where you saved it and how to invoke it.

If anything about my setup is unclear, ask me one or two questions first.

Want to build this yourself?

Join Claude Code Academy: courses, build alongs, and builders turning AI into real systems. And I post a new build on Instagram nearly every day.

Join Claude Code Academy → Follow @aiwithamirthan