Buzz Agent guides

Make Buzz work, one verified layer at a time.

Start with the exact problem. Each guide explains why it happens, what correct behavior looks like, what to do, how to verify it, common failures, and what to try next.

Evidence labels matter: DOCUMENTED is supported by a reliable source, OBSERVED means reproduced in a named environment, and TESTING still needs repeatable proof.

Established cornerstone and governance guides

What Is Block Buzz? Buzz by Block Explained

Learn why Buzz is the official product name, while Block Buzz and Buzz Block are informal search terms, then explore its relay, agents, workflows, setup paths, and current limits.

What Is an Accountable AI Company?

The Buzz Agent Hub operating model and its proof boundaries.

Approval Gates

Decide which actions must stop for human approval.

Permanent Agent Identity

Understand stable per-agent identity and why runtime proof is separate.

Connecting External Capabilities

Govern integrations, repositories, tools, and external agent hands.

Useful companion pages

These pages support the guides without repeating them: orientation, examples, setup flow, and the core Buzz explainer.

Start Here

Use the operator orientation path when you need the right order for learning Buzz.

Agent Snapshots

Explore practical agent examples before you adapt one for your business.

Setup Checklist

Walk through a bounded first-agent checklist instead of improvising the setup sequence.

What Is Buzz?

Confirm the core product model and its limits before you chase an unsupported assumption.

Buzz Updates

See meaningful Buzz changes translated into operator impact, action guidance, and evidence status.

Getting Started

Buzz Agents: The Practical Guide

A practical, source-labeled guide to Buzz agent identity, channels, harnesses, skills, workflows, proof, and recovery.

How to Set Up a Buzz Agent

Set up a Buzz agent with a bounded job, identity, channel access, harness, model, tools, and a real commissioning test.

Configure

Buzz Agent Harnesses: What Runs the Agent

Understand the difference between a Buzz identity and its ACP harness, then verify the live provider, model, prompt, tools, and lifecycle.

Buzz Agent Skills: Install, Scope, and Verify Them

Understand what agent skills are, add only the ones a Buzz job needs, and verify behavior instead of trusting installation output.

Connect

Buzz + Hermes: A Governed Capability Bridge

Use Hermes as an external capability under a Buzz-owned job while preserving ownership, evidence, provider failover, and approval gates.

Coordinate

How Buzz Agents Collaborate

Coordinate humans, agents, channels, handoffs, approvals, and receipts in Buzz without turning the owner into the dispatcher.

Buzz Agent Workflows: Triggers, Gates, and Receipts

Design Buzz workflows around message, reaction, schedule, or webhook triggers with bounded actions, visible receipts, and tested limits.

Keep It Running

Buzz Agent Troubleshooting: A Proof-First Checklist

Diagnose a silent, stale, drifting, or expensive Buzz agent by checking identity, membership, event delivery, harness, provider, model, and artifact proof in order.

Running Buzz Agents Reliably and Continuously

Operate Buzz agents with explicit desired state, heartbeat evidence, bounded recovery, lifecycle tests, and no false online claims.

Control Cost

Buzz Agent Costs and Token Usage

Control Buzz agent cost with job classes, local-first routing, bounded turns, measured provider usage, thresholds, and fail-closed escalation.

FREE newsletter · Buzz Operator Brief

Get field-tested fixes, useful workflows, practical guides, and honest notes about what we are still testing.

By subscribing, you agree to receive the Buzz Operator Brief. Unsubscribe anytime.