Install SintBot
on your Mac

01

Prepare your project Mac

The terminal steps below are for SintBot 1.4.23 on macOS with Node.js 22 or later and Codex or Claude Code. Start with local conversations, or connect 1–32 Feishu China project groups. The desktop App includes Node.js.

• You can start with the automatically created project folder or choose an existing project on this Mac.
• Node.js is version 22 or later.
• Codex or Claude Code (at least one) is installed, signed in and working in your local terminal.
• You can confirm the connection using Feishu on your phone with a team identity.

node --version
codex --version
claude --version

02

Install SintBot in one command

Copy this into your local terminal. The installer checks the environment, installs the package and verifies the result.

curl -fsSL https://sintbot.com/install.sh | bash

curl -fsSL https://sintbot.com/install.sh | bash -s -- --dry-run

SINTBOT="$HOME/.local/share/sintbot/bin/sintbot"
"$SINTBOT" --version
"$SINTBOT" help

SintBot is installed in ~/.local/share/sintbot. Continue below to connect Feishu and your project.

03

Open local setup

Run this command, then follow the Feishu walkthrough for your project, phone confirmation and group. You can also choose “Use on this Mac first”.

"$HOME/.local/share/sintbot/bin/sintbot" start --ui

04

Choose the project on the setup page

By default setup creates ~/Projects/sintbot with a README and binds it to the new group. To connect an existing project, enter its absolute path in the existing-project field; the directory must exist. Do not use your entire home directory as a project. Afterwards, manage project folders and models under Groups · Projects in the console.

05

Everyday start

When setup completes the service is already running; there is nothing else to run. Every later start uses the same command:

"$HOME/.local/share/sintbot/bin/sintbot" start --ui

Keep the terminal open. Settings → Feishu shows the connection status; Groups · Projects manages bindings. Public console access and administrator sign-in are optional.

06

Verify with a real message

Send this in the Feishu group you just connected:

Read the first line of this project’s README.md and reply with only that line.

Compare the reply with the file. This checks the group connection, the runner and the project path together.

Commands and console

Any text message is a task; the reply card appears once the model starts, or after 25 seconds if it is still working without output. While it runs, the card shows how long it has run, what it is doing and the steps already delivered, which fold into a Process section at the end. These commands are answered at once and never take a task slot:

/status

Model, project, busy or idle (steps and current work while busy), and queue.

/skip

Cancel only the current task; queued messages run next and the conversation is kept.

/stop

Stop the current task and cancel queued ones; the conversation is kept.

/new

Start a new conversation, forgetting the previous one, while the group is idle.

/model

Opens a tap-to-pick card: choose the backend, model and thinking depth, then confirm. You can also send /model high, /model <model> or /model <backend>. A model or depth change continues the conversation; a backend change starts a new one. Only while the group is idle.

/hud

Quota of the current model and usage of this conversation.

/issue text

Log feedback for the console.

/help

List commands, with this group's project and model.

Send a command as a message of its own; except for /issue and /model, add no other text. If the Feishu app subscribes to the message-recall event, recalling your own message cancels it.

With --ui, open http://127.0.0.1:4600 on this Mac to see every group, stop tasks, start a new conversation for an idle group, bind new groups, switch the model or project of an idle group, and add or configure models. Set a console password in its settings before enabling LAN access. To add a group without the console, add it to the configuration and restart; it is registered automatically.

Model and thinking depth are chosen per group: on the group card, the Model row picks the backend and model and the next row picks the depth, from lists read live from the local Codex and Claude Code. Choose the model and thinking depth for Codex, Claude and DeepSeek on each group card. DeepSeek connection settings and API key stay in the model library; other third-party endpoints keep their connection model there. After the upgrade, each model card’s existing model and depth move into its groups with the same effective values and no session reset.

Model settings are grouped into model and reasoning, context and compaction, and connection. For Codex, choose the executor default or a custom context window; known limits are checked on save. Compaction also offers default or custom, with numeric inputs shown only when needed. Codex uses a threshold and Claude Code a compaction window; either may compact earlier. Existing explicit global values become independent settings on existing models. Changes apply to the next turn and preserve the session. Capacity details distinguish requested settings from runtime reports.

For Claude Code groups, SintBot can compact the current session between turns when the CLI reports compaction support and the context reaches half of a custom compaction window. It uses a spare concurrency slot and stays silent in the group. A new message interrupts compaction before starting its turn; a window of 0 disables only idle compaction; compaction during a turn follows the executor’s defaults. /hud shows the compaction line and last completion time. Claude subscription quota prefers rate-limit events reported by the executor.

Stop and restart

In the running terminal, press Ctrl + C and wait for the process to exit. Keep the configuration from setup and use the same start command:

SINTBOT="$HOME/.local/share/sintbot/bin/sintbot"
"$SINTBOT" start --ui

Use start for normal restarts. Do not run init again.