LazyCodex
Adds OmO commands, skills, and hooks to Codex so project memory, planning, execution, and verification stay connected.
LazyCodex has a simple role. It connects OmO settings so Codex has a clearer sense of what to remember inside a project, what order to work in, and when to stop and verify.
LazyCodex (LZX) is an OmO Light Edition for Codex. It does not copy all of OmO (oh-my-openagent). Instead, it bundles the Hephaestus-centered flow, commands, skills, hooks, and helper tools that fit Codex well.
After installation, Codex can read project memory, plan before editing, split work into steps, and verify the result with real evidence at the end. The same setup can be recreated later, so a workflow you tuned once is easy to reuse.
Think of it as a tool that gives Codex a work order and a habit of checking its own work.
What You Get
| Feature | In plain words |
|---|---|
| Project memory | Leaves project context in AGENTS.md so Codex does not have to guess from scratch every time. |
| Plan first | Uses $ulw-plan to set the order and scope before code is changed. |
| Plan execution | $start-work follows the plan checklist and keeps the work moving. |
| Evidence check | $ulw-loop and the verification flow look for real evidence, not just “it seems done.” |
| Skills and hooks | Adds specialized flows and automatic checkpoints for review, cleanup, frontend work, programming, and more. |
Remember This First
At first, you only need four steps.
- Create project memory with
$init-deep. - Decide the work order with
$ulw-plan "what to build". - Execute the plan with
$start-work. - Use
$ulw-loop "task"for long work or anything that needs careful verification.
LazyCodex adds rule loading, skills, hooks, model routing, and verification habits to this flow. For details, open only the sidebar pages you need.