DeepSeek Bot turns DeepSeek Harness into a local multi-bot team

A community plugin called DeepSeek Bot adds multiple bots, persistent memory and group chats to DeepSeek Harness, with its data stored on the user’s own computer. It offers Harness users a practical way to experiment with AI teams, though the project is independent of DeepSeek itself.

DeepSeek Watch analysis

What happened

BigGo Finance describes DeepSeek Bot, also called DS Bot, as a plugin for DeepSeek Harness. Each bot can have its own name, role, instructions and model. A main bot called Chief can coordinate other bots in group chats, where users can also call a specific bot directly. The report says an admin bot handles messages by default and brings in others as needed, a design intended to limit the token use of running several bots at once.

The project includes persistent bot and shared-team memory, with tools to view and manage memories. According to the report, memory is saved as Markdown files, and the plugin uses summarisation and local search to handle long conversations. It also provides token-use statistics by time, bot and model. The project is open source under the Apache License 2.0 and stores data locally; the report notes that uninstalling the plugin does not remove that data.

Why it matters

This is a community-built extension, not an official DeepSeek multi-agent product. Its appeal is less about making a grand claim for AI agents and more about making a team of them testable without a separate server deployment. The admin-and-mention setup also gives users a way to decide when another bot joins in, rather than paying for the whole virtual committee to answer every message.

Persistent memory and local storage bring practical responsibilities with them. Users should know where the saved files live and remove them separately if they no longer want them. BigGo describes the project’s features, but provides no comparative test of memory quality, reliability or token savings.

Our read

DeepSeek Bot is a useful-looking piece of ecosystem plumbing: several bots, shared and individual memory, and a local-first setup in one plugin. The most persuasive detail is its on-demand group-chat design, which tries to make collaboration useful without turning every prompt into a roll-call. Treat it as an independent community project to explore, not as an official DeepSeek release or a proven productivity upgrade.

What to watch

• Whether the project publishes clear setup and data-removal guidance for its local memory files.
• How well the coordinator and group-chat approach works in real use, including token costs.
• Whether the project expands model support as its author has suggested.

Discussion spark: Would you trust a local multi-bot setup to keep useful long-term memory, or would you rather give up some convenience to keep each AI session separate?

not affiliated with or endorsed by DeepSeek