Block, the company behind Square and Cash App, shipped Buzz this week as a free, open-source alternative to running separate paid tools for team chat, code review, and workflow automation, and its distinguishing design choice is treating AI agents as full participants rather than bolted-on integrations. Buzz runs on the Nostr protocol, which means every participant, human or AI agent, gets its own portable cryptographic keypair as an identity rather than a platform-issued account. That sounds like a detail specific to the underlying protocol, but it has a concrete practical payoff: every action an agent takes inside Buzz, a chat message, a code patch, a workflow step, an approval, gets recorded as a signed, append-only event tied to that agent's actual identity, producing an audit log that survives independently of Buzz itself and can't be quietly altered after the fact. This addresses a problem that's become increasingly real as teams give coding agents write access to shared repositories and communication channels: once an agent can open pull requests, review code, and post in the same channels as human teammates, you need a reliable way to answer which agent did this and whether the record of what it did can be trusted, and a signed event log is a more durable answer to that than relying on whatever internal logs a single SaaS vendor happens to keep. Buzz ships with pre-built harnesses for three existing coding agents, Block's own Goose framework, OpenAI's Codex, and Anthropic's Claude Code, wired in through the Agent Client Protocol, an open standard for connecting agents to developer tools, so teams aren't locked into a single agent vendor to get these guarantees. For engineering teams already running multiple agents across chat, code review, and automation, this is worth a look specifically because it tackles the identity and accountability layer that most agent-enabled tools currently treat as an afterthought.