Open Tag: The Open‑Source Counterpart That Quickly Matches Claude Tag’s AI Agent Features

Anthropic recently launched Claude Tag, an AI‑agent that can be summoned in Slack to summarise conversations, organise pull‑requests, and prioritise tasks; the feature quickly became a productivity tool for many teams.

Within a few days, Atai Barkai, a CopilotKit developer, released Open Tag as an open‑source alternative. Unlike Claude Tag, which is tied to the Claude model, Open Tag can connect to any large language model or agent framework and allows developers to fully customise the agent logic. Its core capabilities—generative UI interaction, streaming responses, human‑in‑the‑loop approval, and complete thread‑context retrieval—cover all of Claude Tag’s functions while adding extra customisation space.

Open Tag currently supports Slack and Microsoft Teams, with plans to add Discord, Google Chat, Telegram, and WhatsApp. The key advantage is that the same AI agent can run on all these platforms without code changes, eliminating the need for per‑IM adaptations—a task that developers describe as a "messy, hard‑to‑do" chore.

The project’s initial rollout caused a minor controversy: the early access link was a Google Form and the source code was not yet public, prompting accusations of false open‑source claims. Atai clarified that the project follows the CopilotKit and AG‑UI open‑source licenses and later published the full GitHub repository, which immediately quelled the criticism.

The repository (https://github.com/CopilotKit/OpenTag) is advertised as deployable within minutes. Early adopters report plans to stealthily install it in company Slack workspaces, to extend personal tools they previously built, and they raise practical questions about channel‑specific permissions, per‑channel context separation, and using their own Claude subscriptions; the developers continue to respond.

This rapid iteration illustrates how closed‑source AI tools can be eclipsed by open‑source equivalents that not only replicate functionality but also add flexibility. For teams with custom requirements, an open‑source solution may be preferable despite the additional time needed for deployment and debugging.