Skills
Specialist playbooks LazyCodex loads on top of the command pillars.
LazyCodex packages OmO (oh-my-openagent) as the Codex agent harness. On top of the command pillars ($ulw-plan, $start-work, $ulw-loop, /init-deep), the OmO plugin installs a set of skills: reusable instruction packs that turn a focused worker into a domain specialist.
Skills are invoked from inside a Codex session. The discipline agents (Sisyphus, Hephaestus, Oracle, Librarian) load the skill whose domain overlaps the task, then follow its playbook.
Installed skills
| Skill | Use it for |
|---|---|
review-work | Multi-angle post-implementation review |
remove-ai-slops | Behavior-preserving cleanup of AI-looking code 1 |
debugging | Systematic root-cause debugging |
refactor | Test-verified refactoring, including structural search and rewrite via ast_grep_search / ast_grep_replace |
frontend-ui-ux | Designed UI work instead of generic layout filling |
programming | Strict TypeScript, Rust, Python, or Go discipline |
LSP | Diagnostics, definitions, references, symbols, and renames |
rules | Project instructions from AGENTS, rules, and instruction files |
comment-checker | Feedback after edit-like operations |
ultragoal | Goal-locked, evidence-driven completion |
How skills fit in
- Command pillars own the loop: planning, execution, and verified completion.
- Skills add specialist judgment around those pillars: review, cleanup, UI craft, language discipline, and code intelligence.
- A skill is loaded for the task at hand and unloaded after, so agent prompts stay lean.
Where skills live
OmO can load skills from project and user locations such as .opencode/skills, ~/.config/opencode/skills, .claude/skills, .agents/skills, and ~/.agents/skills.
Per-skill docs
review-work
remove-ai-slops
frontend-ui-ux
debugging
refactor
ultragoal
Command pillars
$ulw-* and /init-deep workflows skills run on top of.Footnotes
-
ai-slop-removeris the single-file worker thatremove-ai-slopsinvokes in parallel — it is not a separate user-facing skill. ↩