What does this skill do, and when should you use it?
This skill is an authoritative guide to project architecture and file structures for Electron multi-process applications. It offers a decision tree, process boundary rules, naming conventions, and structural rules to determine where new code belongs, avoiding runtime crashes from cross-process imports. It references three detailed docs covering renderer layer, main process and shared layer, and project root/monorepo layout. Originating from the AionUi desktop app repository, it is open-source under Apache-2.0.
Provides a decision tree to route new code to the correct directory (e.g., UI components to renderer, IPC handlers to bridge, AI platform connections to agent). Defines hard process boundary rules—which APIs each process can use (main, renderer, worker, preload). Gives naming conventions for directories (PascalCase for React components, lowercase for categorical dirs) and files. Includes structural rules like directory size limits and test file mapping, plus a quick checklist.
- When adding a new React hook, quickly decide whether to place it under renderer/hooks or a shared location.
- Reviewing a PR to catch potential cross-process imports that would crash at runtime.
- Converting a single-file component into a directory when it gains private sub-components.
- Onboarding a new developer to the codebase structure and conventions.
- Setting up tests for new source files following the mapping table.
How do you install this skill?
- Referenced documents may become out of sync with the codebase; verify current directory structures before use.
- The skill is in English; for Chinese users, provide a Chinese version or ensure Chinese prompts trigger correctly.
- The skill lacks maintenance and update information; users should monitor repository versions and changes to avoid using outdated structures.
The skill is part of the AionUi repository. To use it, clone the repo or copy the .claude/skills/architecture/ folder into your project's skills directory (e.g., .claude/skills/). Since it's a single skill, you can also copy just that folder.
tmp="$(mktemp -d)"
git clone --depth 1 https://github.com/iOfficeAI/AionUi.git "$tmp"
mkdir -p ~/.claude/skills
cp -R "$tmp/.claude/skills/architecture" ~/.claude/skills/
rm -rf "$tmp"Generated from the source repository and skill path; it copies only this skill's folder. If the author's install steps above differ, follow those first. To scope it to one project, replace ~/.claude/skills with that project's .claude/skills.
How do you use this skill?
Invoke the 'architecture' skill in Claude Code or a compatible agent. It will provide guidance based on its decision tree and rules. The skill is a knowledge artifact, not a script; you manually apply its conventions to place code and review structure.
What are this skill's strengths and limitations?
- Clear decision tree simplifies code placement
- Hard process boundaries prevent crashes
- Detailed naming conventions ensure consistency
- Test mapping helps maintain test structure
- Practical focus on Electron multi-process projects
- No automated enforcement; relies on manual adherence
- Reference paths are specific to AionUi repo, not generic
- No examples or walkthroughs included
- Assumes familiarity with Electron and monorepo layouts
How does this skill compare with similar options?
Side by side with related skills; every score comes from the same FSRS standard.
| Skill | FS score | Stars | Last updated | License |
|---|---|---|---|---|
| AionUi Architecture Conventions this page | 57 · Use with care | ★ 33k | 1mo ago | Apache-2.0 |
| AGENTS.md Context Builder ✓ Microsoft · Official | 54 · Use with care | ★ 2.7k | 3mo ago | MIT |
| Expo App Structure Guide | 57 · Use with care | ★ 2.7k | 6d ago | MIT |
| VS Code Unit Test Runner ✓ Microsoft · Official | 55 · Use with care | ★ 193k | 3d ago | MIT |
| OpenChamber Desktop Shell Skill | 54 · Use with care | ★ 11k | 3d ago | MIT |
How did FollowSkills review this skill?
The skill only provides code placement and structure guidance, involving no file system operations, network calls, or sensitive data handling, hence requiring minimal permissions. It clearly declares process boundary rules and directs IPC through preload bridging, aligning with least privilege. However, the actual consistency of the referenced documents with the codebase is unverified, and boundary rules could lead to misuse of cross-process imports, hence deduction.
The skill is self-consistent with a clear decision tree and boundary rules, without contradictions. However, all paths and directory structures are based on referenced documents whose alignment with the codebase is unverified. Static review cannot execute, and failure feedback mechanisms (i.e., providing diagnosable error messages) are unclear. Thus score is below happy path.
The skill provides clear triggers (creating new files, deciding code placement, etc.) and clear scenario classifications, but non-fit boundaries are not explicitly stated. The skill is specific to this project structure and may be less applicable to users unfamiliar with it. Also, the skill is in English, which may hinder Chinese users, but its functionality entirely depends on local codebase structure and not on overseas networks, so no deduction for that. Triggers could still be refined.
The skill uses layered reference docs (process.md, renderer.md, project-layout.md) with good information architecture, decision tree, and checklists. However, installation/dependency notes, changelog, versioning, and maintenance ownership are missing. License and provenance are clear in repository context but not in the skill itself. Update path is unclear, hence deduction.
The skill can help users determine code placement, but whether it fully accomplishes the user's task depends on its accuracy and user understanding. Static review cannot verify actual output, and the skill is a guide, not an automated tool, so marginal value is limited. Hence score below core task completion.
The content is based on project structure, but static review cannot execute or independently verify. The repo contains tests and build workflows, but not for this skill itself, so evidence coverage is limited. Author claims and repo structure are auditable, but independent verification of the skill is lacking. Hence score capped at 5.
Open a dimension to read why it scored that way
Evidence confidence：Low — Mostly static review, author material or a limited demo; useful for discovery, not high-risk decisions.
See the full review method →