--- name: log-instagram-dm-reply description: "Reading and answering an Instagram DM. A log play (docs/useCases/instagram/dmReply.md): Read what someone wrote to the account on Instagram and answer it in the owner's voice, with text or a meme, on the account's own session through vendor/instagrapi. No browser, ever." when_to_use: "Use when the owner asks for this, in words like: \"read my instagram dms\", \"what did say on instagram\", \"reply to on instagram\", \"send a meme on instagram\"" metadata: summary: "Reading and answering an Instagram DM" --- # Reading and answering an Instagram DM This is the log play `instagram/dmReply`. Before anything else, call the log MCP tool `getGuide({ useCase: "instagram/dmReply" })`: it returns this page and makes it the session's play, so every log tool's answer says which step you are on and a send with no play loaded is refused (rules 157, 158). Then follow the steps in order. Draft every message from that person's own thread and the owner's real replies (`voice` with no text, `search`), never from memory, and never send two that read alike (`sameAsSent`). Read what someone wrote to the account on Instagram and answer it in the owner's voice, with text or a meme, on the account's own session through vendor/instagrapi. No browser, ever. How the owner asks for it: "read my instagram dms", "what did say on instagram", "reply to on instagram", "send a meme on instagram" ## Run it from a fresh context Say: **"go on docs/useCases/instagram/dmReply.md for "**. The agent reads this page and runs the steps. ## The steps | # | Step | Tool | Done when | | --- | --- | --- | --- | | 1 | Session check | `instagramSessionCheck` { account } | `loggedIn` as the account's handle | | 2 | The inbox, or the thread with them | `instagramDms` { account }, then `instagramDms` { account, with: "" } | their latest message and what came before it | | 3 | Pull how the owner writes to them, then draft only from facts the owner gave | `voice` { to: "", with: "", context } (no text) | a draft that answers what they said | | 4 | For a meme, find one that fits the moment, ranked by the owner's taste, and save its picture | `memelordSearch` { query: [""], method: "fulltext", type: "meme" } | a template that fits, saved as a JPEG | | 5 | Gate the draft or the meme | `voice` { to, with, context, text: "" } | it passes | | 6 | Send it: text one message per line, a meme as `photo` | `instagramSendLive` { account, with: "", text and/or photo: "", approvedBy: "", confirmLive: true } | Instagram answers each message with its id | | 7 | Tell the owner what they said and what went out | the reply to the owner | both quoted with their times | New messages arrive by themselves: the inbox hook keeps `log instagram watch --inbox` open for every signed-in Instagram account on Instagram's realtime connection and puts each new DM in front of the agent (`>>> NEW INSTAGRAM MESSAGE`). ## What the runs taught - **Never the browser.** On 2026-11-04 a meme went out through the account's [AdsPower](https://www.adspower.com/share/pD25Iu) browser and the owner said that is not the way: read, send and watch are the engine's. - **A meme goes as the picture.** [Memelord](https://www.memelord.com/sign-up?ref=kfwut9wc) answers with a media URL; save it as a JPEG and send it as `photo`, never the link. - **Pick from the top of the taste ranking, with judgement.** A template built on a stereotype is skipped even when it ranks first. > Some links in this play carry ListeningKit's referral or affiliate credit (a `ref=` or `aff=` code). They cost the owner nothing extra; use the plain address without the code if you prefer.