Buzz: A Self-Hosted Workspace for Collaboration Between Humans and AI Agents

block released the self-hosted workspace Buzz on GitHub on July 24, 2026. Built on Nostr protocol relays, Buzz provides a collaborative environment where humans and AI agents can share the same room to work together on tasks such as software development, reviews, and workflow executions.

Overview

Buzz is designed as a workspace where “humans and AI agents collaborate in the same room.” Its standard setup involves a single community configuration with a self-hosted relay. While operators can host multiple communities across various domains and subdomains, client-side rules remain consistent. A URL serves as the workspace’s ID, and all tenant states under it are confined locally within the community.

Design Based on Nostr

At the core of Buzz is the Nostr relay. Every message, reaction, workflow step, review approval, and Git event is recorded as a signed event in a single log. Whether authored by a human or a process, the same data structure, ID model, and audit trail are maintained. While it functions as a team-oriented workspace on the surface, its internal implementation is an event log composed of numerous Rust crates.

This architecture ensures that even in multi-tenant deployments sharing Postgres, Redis, and object storage backends, each community maintains semantically independent boundaries. The philosophy underlying Buzz is that a single community can replace traditional team workspaces, with all events fitting into a unified search index.

Uniform Model for Agents and Humans

What sets Buzz apart from conventional AI development tools is its treatment of agents as “members.” Agents are granted the same privileges and interface access as human teammates. They can open repositories, submit patches, review code, execute workflows, edit canvases, orchestrate other agents, participate in voice huddles, create channels, and summon relevant stakeholders. The only distinction lies in their key pairs.

Agent actions are not scoped by permission flags but by IDs, recorded in the same audit trail and event log. Agents can do more than just converse in chat—they actively operate the workspace itself. Channels, canvases, workflows, huddles—agents can interact with all these interfaces on par with humans.

For example, an agent can search six months of project history to answer a question and present the relevant thread. Without granting full repository access, agents can triage bugs. By converting feature branches into rooms, they manage patches, CI, reviews, and merge decisions within the same channel, ensuring that the rationale for code existence is documented as part of the channel itself.

One Community, One Event Log

“One community. One identity model. One event log.” This phrase encapsulates Buzz’s design philosophy. Humans, agents, workflows, and repositories all communicate using the same protocol, sign with identical keys, and are stored in the same search index.

In traditional development tools, chat platforms for humans, CI/CD pipelines, code review tools, and project management systems typically exist as separate entities. Agents are often limited to bot accounts with restricted permissions. Buzz integrates all these elements into a single event log, significantly enhancing auditability and searchability. Additionally, knowing just one API endpoint enables all operations within Buzz.

Comparison with Competitors and Buzz’s

Positioning

The emergence of Buzz can be seen as part of a broader movement to fully integrate AI agents into development workflows. Unlike approaches that add AI bots to existing tools like Slack or Discord, Buzz redesigns the workspace itself as an event log.

As competition in agent technology heats up—illustrated by Anthropic’s announcement of Fable 5’s return—self-hosted collaboration platforms like Buzz offer organizations a way to operate humans and agents on a unified foundation while retaining data sovereignty. Its release under the Apache 2.0 license also enables the community to fork and customize the platform as needed.

Editorial Opinion

In the short term, Buzz presents a compelling option for organizations looking to adopt AI agents in their development teams while avoiding dependence on external cloud services. Its foundation on the Nostr protocol also promises interoperability with the existing Nostr ecosystem. However, deploying and managing Buzz requires knowledge of Nostr and Rust, which could pose a significant adoption barrier.

In the long term, the model of humans and agents operating on the same event log has the potential to revolutionize software development audits and traceability. By recording every development process as a signed event log, it becomes possible to fully reconstruct the reasons behind code changes and decision-making processes. Whether this approach becomes an industry standard depends on the ease of operation and the growth of its community.

Granting agents repository manipulation and workflow execution rights surpasses the traditional concept of bots but also raises security concerns. Proper key pair management and ID-based scoping of agent permissions will be critical challenges in real-world implementations.

References

• “block /

  buzz", by **block** — GitHub Trending, 2026-07-24 (ARR)

• Source URL: https://github.com/block/buzz