openTag

openTag is an open-source AI coworker from first-tree.ai. You mention it in a Slack or Lark thread, it hands the task to a coding agent running on your own computer, and the reply comes back in the same thread. It is licensed under Apache 2.0 and can be self-hosted.

What It Is

opentag connects a messaging workspace to AI agents that run locally on a Mac or Linux computer. Supported agent runtimes are Codex, Claude Code, and Pi, and the site lists Cursor, Kimi, OpenCode, Z.ai, and DeepSeek as coming soon. Agents use your own model subscription or API key, so there is no single-provider lock-in.

How the Workflow Runs

Setup has three steps: create an agent and choose its runtime, connect your computer by pasting a connection command into a terminal, and connect Slack or Lark. After that, one agent can serve a channel and anyone in it can mention it. Agents can also delegate work to each other and check on each other's progress. The project says its servers do not store conversations, code, or memory by default, and that connectors use your own credentials.

Team Memory and Connectors

Shared team memory is kept in a Context Tree, a file-based, git-backed markdown repository hosted in your own GitHub organization. The site says agents can connect to over 1000 services, including GitHub, Notion, Linear, Google Drive, Jira, and Figma. Teams and Discord support is listed as on the roadmap.