ALO

A workflow engine for coordinating coding agents, commands, reviews, and retries, with saved state so a long-running task can resume instead of starting over.

Go
YAML
Git

Conceptual workflow for ALO: Plan, then Execute, then Review. — Open full-size screenshot or illustration

Make complex agent workflows repeatable.

A multi-step engineering task should have an explicit route through planning, implementation, and review. ALO expresses that route in YAML, including loops, conditions, command steps, and approval gates.

How it works

Run state survives interruption. Budget limits and turn caps put boundaries around execution, while worktree isolation separates changes from the main checkout. Watch, replay, and diff surfaces make a run inspectable.

Workflow example

Define a workflow with an implementation step and a validation command.
Run it against an isolated Git worktree and inspect the live state.
If interrupted, resume the saved run; review the result before merging.

Project scope

The local CLI requires a supported external agent, its authentication, and a working Git environment. This page illustrates the workflow; it does not run agents in your browser.