From a71f167428fbad5f2bd515a831318f77a9a1809d Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Lu=C3=ADs=20Serralheiro?= Date: Sat, 20 Jun 2026 12:02:18 +0100 Subject: [PATCH] Add AFK integration Co-Authored-By: AFK using gpt-5.5 --- 4_integrations/afk.mdx | 94 ++++++++++++++++++++++++++++++++++++++++ 4_integrations/meta.json | 1 + 2 files changed, 95 insertions(+) create mode 100644 4_integrations/afk.mdx diff --git a/4_integrations/afk.mdx b/4_integrations/afk.mdx new file mode 100644 index 0000000..ae890e0 --- /dev/null +++ b/4_integrations/afk.mdx @@ -0,0 +1,94 @@ +--- +title: AFK +description: Use AFK with LM Studio +index: 5 +--- + +import { Step, Steps } from "fumadocs-ui/components/steps"; + +AFK can talk to LM Studio through LM Studio's OpenAI-compatible local server. Use AFK's browser UI to create a Local model connection, then choose your LM Studio model when starting an AFK coding-agent session. + + +Have a powerful LLM rig? Run LM Studio on that machine and connect an AFK daemon that can reach it. In AFK, `localhost` is resolved from the daemon machine. + + +## Setup + + + +

Start LM Studio's local server

+ + Make sure LM Studio is running as a server (default port `1234`). + + You can start it from the app, or from the terminal with `lms`: + + ```bash + lms server start --port 1234 + ``` + + + +

Create or sign in to AFK

+ + Open [afk.mooglest.com](https://afk.mooglest.com) and create an account or sign in. + + + +

Install and connect an AFK daemon

+ + AFK sessions run through a daemon that has access to your projects and can reach LM Studio. + + In AFK: + + 1. Open **Account → API Keys**. + 2. Create a daemon token. + 3. Follow the installation command shown in the app to install and connect the daemon. + 4. Confirm your machine appears as a connected daemon in the browser. + + If LM Studio is running on the same machine as the daemon, use `http://localhost:1234/v1` as the base URL. If LM Studio is running on a different machine, use a URL the daemon machine can reach. + + + +

Add a Local model connection in AFK

+ + In AFK: + + 1. Open **Account → LLM**. + 2. Click **Add connection** and choose **Local model**. + 3. Set **Base URL** to: + + ```text + http://localhost:1234/v1 + ``` + + 4. If Require Authentication is enabled in LM Studio, paste your LM Studio API token as the API key. + 5. Bind the connection to the daemon machine that can reach LM Studio. + 6. Save or test the connection. + + + +

Start an AFK session with an LM Studio model

+ + Click **New session** in AFK, then: + + 1. Select the connected daemon and project directory. + 2. Choose the LM Studio Local model connection. + 3. Select or type the model identifier exposed by LM Studio. + 4. Choose a permission mode and enter the coding task. + + AFK will route the session's model requests to LM Studio. + + + Use a model and context length large enough for coding-agent work. Agent sessions can consume a lot of context, especially when reading files and using tools. + + + + +## Troubleshooting + +- **Connection fails:** Confirm LM Studio's server is running and reachable from the AFK daemon machine. +- **`localhost` does not work:** In AFK, `localhost` means the daemon machine, not the hosted AFK web app. Use the daemon-reachable host or IP if LM Studio is on another machine. +- **Authentication errors:** If Require Authentication is enabled in LM Studio, create an LM Studio API token and paste it into the AFK Local model connection. +- **Model is missing:** Load a model in LM Studio and try testing the AFK connection again. You can also type the model identifier manually when starting a session. + +If you're running into trouble, hop onto the [LM Studio Discord](https://discord.gg/lmstudio). diff --git a/4_integrations/meta.json b/4_integrations/meta.json index da9587e..df72910 100644 --- a/4_integrations/meta.json +++ b/4_integrations/meta.json @@ -8,6 +8,7 @@ "codex", "hermes", "openclaw", + "afk", "---MCP Integrations---", "...1_mcp-remote" ]