openTag for Run Ai Agents Locally
Run AI agents locally with openTag
openTag fits readers who want a chat-based agent workflow, but want the actual work to happen on their own computer. It is an open source AI coworker for Slack and Lark: you mention it in a thread, it passes the task to an agent running on your machine, and the reply comes back in the same conversation. The product says you can use your own AI subscription or API key, so the agent is not tied to one provider.
The local part is the key distinction. openTag describes the agent as running on your computer, with messages and results passing through openTag. It also says, by default, its servers do not store your conversations, code, or memory. For a local setup, that means the computer you connect is where the agent executes the task, rather than a hosted workspace running everything on the vendor side.
The documented setup is straightforward. openTag shows three steps: create an agent, prepare the computer, and connect messaging. The computer side uses a connection command from the app, and the messaging side connects the agent to Slack or Lark. The site also says one agent can serve a channel, so several people in that workspace can mention the same agent.
openTag also includes shared team memory through Context Tree, a file-based, git-backed memory layer hosted in the team’s GitHub organization. That matters for local agent work because the agent is not just answering one-off prompts; it can use team context that stays in files your organization controls. The product also points to connectors for more than 1,000 services, including GitHub, Notion, Google Drive, Jira, and Figma, so the local agent can work with context from outside chat as well.
The supported environment is limited to what openTag names: Slack and Lark for messaging, and Mac or Linux for the machine that runs the agent. It also says more agents are on the way, so the current local setup is tied to the agent and app options the product already lists.