Homeboy Model
Homeboy starts with a component and keeps the command model consistent across where that component runs.
Component
A component is the buildable, testable, reviewable unit. Portable component config usually lives in homeboy.json at the repository root.
Extension
Extensions provide ecosystem-specific behavior while Homeboy core stays generic. For example, an extension can teach Homeboy how to run Cargo, WP-CLI, Node.js, GitHub releases, or platform-specific audit rules.
Command Surface
Homeboy exposes one operational surface for local developers, CI, scheduled jobs, and coding agents:
homeboy review audit
homeboy review lint
homeboy review test
homeboy review
homeboy bench
homeboy trace
homeboy releaseEvidence
Commands print human-readable output and can write structured JSON with --output. Longer workflows can persist runs and artifacts for later inspection with homeboy runs.
Runners
Runners let Homeboy route hot or remote-capable commands to another execution environment while preserving the same command contract.
Headless Orchestration
Agent tasks, loops, controllers, runs, and artifacts make Homeboy useful without a chat session or interactive shell. Durable ids let terminals, CI jobs, runners, scheduled tasks, and chat bridges observe and resume the same work.
Operations
Projects, servers, and fleets add remote operations for deployments, SSH, logs, files, databases, and fleet fan-out. These are useful for configured environments but are not required for the local quality loop.