From oma
Project harness initialization — Generate AGENTS.md, ARCHITECTURE.md, docs/ structure
How this skill is triggered — by the user, by Claude, or both
Slash command
/oma:deepinitThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
Response language follows the `language` setting in `.agents/config/user-preferences.yaml`.
Response language follows the language setting in .agents/config/user-preferences.yaml.
.gitignore; skip framework-generated cross-platform build directories (android, ios, macos, linux, windows, web)5-part analysis using Glob, Grep, Read:
Report findings before proceeding.
Content: domain map, package layering, dependency direction, data flow, key integration points. Include mermaid diagrams where helpful.
Rules: under 200 lines; topology and relationships only, not implementation details.
Generate only files relevant and discoverable from the codebase.
Directory structure:
docs/
├── design-docs/
│ ├── index.md ← catalogue with status (draft/verified/superseded)
│ ├── core-beliefs.md ← agent-first operating principles
│ └── {decision-name}.md
├── exec-plans/
│ ├── active/
│ ├── completed/
│ └── tech-debt-tracker.md
├── product-specs/
│ ├── index.md
│ └── {feature-name}.md
├── references/
│ └── {library}-llms.txt
├── generated/
├── DESIGN.md, FRONTEND.md, PLANS.md
├── QUALITY-SCORE.md, RELIABILITY.md, SECURITY.md
└── CODE-REVIEW.md
Domain docs generation (root-level in docs/):
| File | When |
|---|---|
DESIGN.md | Project has UI/design system |
FRONTEND.md | Project has frontend |
PLANS.md | Always |
QUALITY-SCORE.md | Always |
RELIABILITY.md | Has backend/infra |
SECURITY.md | Has auth/data handling |
CODE-REVIEW.md | Always |
Write only what was discovered. Mark uncertain patterns <!-- TODO: confirm this rule -->.
~100 lines. Sections:
<!-- MANUAL: Notes below this line are preserved on regeneration -->Rules: no file listings; every line points somewhere or states a rule; no dead links.
Only at package/app boundaries in monorepos. Max 50 lines each.
Template:
<!-- Parent: ../AGENTS.md -->
# {Package/App Name}
> {One-line purpose}
## Constraints
{Rules specific to this boundary}
## Working Here
{How to correctly add/modify code in this area}
## Dependencies
- Depends on: {internal deps}
- Depended on by: {reverse deps}
<!-- MANUAL: Notes below this line are preserved on regeneration -->
If existing AGENTS.md found:
<!-- MANUAL: --> and <!-- TODO: --> blocks exactly<!-- REVIEW: this rule may be outdated -->active/ to completed/Validation checklist:
<!-- Parent: --> references in boundary AGENTS.md resolve correctlydocs/design-docs/index.md lists all design docs that existdocs/product-specs/index.md lists all product specs that existReport validation results to the user.
$ARGUMENTS
npx claudepluginhub kyongsik-yoon/oh-my-agent --plugin omaCreates structured, bite-sized implementation plans from specs or requirements before writing code. Useful for breaking down multi-step tasks into testable steps with file structure and task boundaries.
2plugins reuse this skill
First indexed Jul 17, 2026