> For the complete documentation index, see [llms.txt](https://djthorpe.gitbook.io/go-llm/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://djthorpe.gitbook.io/go-llm/messaging.md). # Messaging Connect chats on messaging services to a conversation, so you can talk to your assistant from your phone. Telegram is the first. Slack ([#249](https://github.com/mutablelogic/go-llm/issues/249)) and email ([#305](https://github.com/mutablelogic/go-llm/issues/305)) are planned. ## Telegram 1. Make a bot with [BotFather](https://t.me/BotFather), and set its token as `TELEGRAM_BOT_TOKEN`. 2. In a conversation whose model has the `messaging` tools, ask to talk on Telegram. You get a link, valid once for 15 minutes, which the web UI also shows as a QR code. 3. Open the link, or scan the code, and press Start. ```bash llm session-create --model ollama:llama3.2 --tools "messaging:*" llm chat "Let's talk on Telegram" ``` ## Channels * **Both ways:** what you say on Telegram goes to the conversation. Every reply, and every task's outcome worth telling you, goes to all its channels. * **Managed by the assistant:** it connects, lists, sends to and disconnects chats, with the `messaging` tools. * **More than one:** a conversation can have several chats, such as yours and a group's. A chat belongs to one conversation, so connecting it again moves it. * **Groups:** everyone in a connected group can talk to the assistant, and see what it remembers about you. Only connect groups whose people you'd tell. * **Who said it:** each message says where it came from, and who sent it. * **Ignored:** messages from chats which aren't connected are ignored. ## Running the bot * **No public address needed:** the server reads the bot's messages by long polling. * **One server per bot:** don't share a bot's token between servers. * **Progress:** while a reply takes a while, the chat shows what's happening, in one message, which is deleted when the reply arrives. ## Replies * **Formatting:** Markdown is converted to Telegram's. Long replies are split between paragraphs. * **Files:** pictures and speech a reply makes are sent with it. Images go as photos, Ogg audio as voice messages, MP3 and M4A as tracks, and anything else as documents, up to 50 MB. ## Voice notes To talk by voice, the assistant needs a model that speaks, and one that transcribes. With the compose file's `ollama` and `whisper` profiles, these run on your own machine: ```bash llm model-load ollama:phonon:latest # speaks llm model-load gowhisper:ggml-medium-q5_0 # transcribes ``` * **Tell the assistant which to use,** such as "speak with `ollama:phonon:latest` in Alfie's voice, and transcribe with `gowhisper:ggml-medium-q5_0`". * **Speaking:** on Telegram, replies are sent as Ogg voice notes. Models that can't make them need `opusenc`, which the Docker image has. * **Transcribing:** `ggml-medium-q5_0` runs on small machines, but can mishear names. `ggml-large-v3-turbo-q5_0` hears them better, on a faster machine. ## Files you send * **Kept with the conversation:** photos, documents, voice messages, audio and video become the conversation's files. A caption is the message's text. * **Albums** arrive as one message. * **Over 20 MB:** bots can't download files this large, so the assistant is told it didn't get them. --- # Agent Instructions This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com. ## Querying This Documentation If you need additional information that is not directly available in this page, you can query the documentation by asking a question. Perform an HTTP GET request on the following URL with the `ask` and `goal` query parameters: ``` GET https://djthorpe.gitbook.io/go-llm/messaging.md?ask=&goal= ``` `ask` is the immediate question: it should be specific, self-contained, and written in natural language. `goal` is what the user is ultimately trying to achieve, the reason they need the answer. Sharing it helps GitBook give you a better, more relevant answer. A goal is most helpful when it describes the outcome the user wants rather than restating the question. For example, with `ask=how do I create an API token`, a goal like `automate deployments from our CI pipeline` lets GitBook tailor the answer to that use case. The response will contain a direct answer to the question and relevant excerpts and sources from the documentation. Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.