search

Documentation Home

GitHub-native AI engineering CLI. Turn issues into shipped code.

Locus is the unified AI engineering interface for GitHub teams: one CLI to plan, execute, review, and ship code across Claude and Codex.

hashtag
Getting Started

If you are new to Locus, follow these pages in order:

  1. Installation -- install Locus, gh, and one AI provider CLI.

  2. Sandboxing Setup -- set up Docker-based isolation for safe AI execution.

  3. Quickstart -- run one complete end-to-end workflow.

You can reach a working flow by following only those three pages.

hashtag
Who This Is For

  • Engineering teams already using GitHub Issues, Milestones, labels, and PRs as their delivery workflow.

  • Teams that want one operational interface across Claude and Codex instead of provider-specific process.

  • Developers who want built-in planning, execution, review, iteration, and status workflows in one CLI.

  • Teams that want full-auto execution with Docker sandbox isolation and GitHub-native auditability.

hashtag
What Locus Is Not

  • Not a replacement for GitHub. GitHub is the system of record.

  • Not a hosted SaaS backend. Locus runs locally and uses gh for GitHub operations.

  • Not "just model access". Locus provides orchestration workflows on top of provider CLIs.

  • Not one-click magic. Automation is explicit and configurable (autoLabel, autoPR, run --resume).

hashtag
Core Strengths

  1. Unified interface across AI clients -- switch models and providers without changing your workflow commands.

  2. GitHub-native operational memory -- issues, milestones, labels, and PRs are the entire data plane. No external database.

  3. Built-in orchestration tools -- plan, run, review, iterate, status, and logs go beyond what raw provider CLIs offer.

  4. Safe automation via auto-approval mode -- full-auto execution with Docker sandbox isolation, resumable runs, and configurable safeguards.

hashtag
What You Will Learn Next

The Quickstart walks through:

  1. Switching between Claude and Codex using the same command surface.

  2. Creating and executing GitHub-native sprint work.

  3. Running built-in plan -> run -> review -> iterate workflows.

  4. Enabling automation settings and resuming interrupted execution.

After that, dive deeper into:

  1. How Locus Works

  2. Execution Model

  3. Sprints and Issues

  4. Built-In Tools

  5. Auto-Approval Mode

NextInstallation

Last updated