Reusable AI agent skills, decision frameworks, and improvement roadmaps.
Creates requirement-driven architecture and stack recommendations. It prefers the simplest viable design, evaluates technology against project constraints, explains trade-offs, applies SOLID pragmatically, and produces concise Architecture Decision Records when needed.
Location: skills/software-architecture-advisor
Install guide: skills/software-architecture-advisor/INSTALL.md
Plans documentation strategy, authors feature analyses/specifications, handles documentation-to-code synchronization, and designs optional cross-session agent memory. For repository-wide documentation structure and splitting rules, use Context Engineering.
Location: skills/doc-strategy-engineer
Source of truth for repository documentation structure and document-splitting policy. Establishes routing/indexes, audits retrieval and authority, and executes justified structural migrations with a preservation map. Documents over 150 lines trigger review, not automatic splitting; the line thresholds are local conventions, not hard limits.
Location: skills/context-engineering
Use Context Engineering for large-document review/splitting, index/routing repair, and structure audits; use Documentation Strategy Engineer for feature documentation, documentation strategy proposals, and agent-memory lifecycle. For mixed tasks, use the latter to author content and the former's structural policy and preservation workflow for any restructuring. A justified in-scope split needs no extra approval based on file count; owner decisions about changed meaning, security/product/architecture, or scope expansion still require approval.
Install with python scripts/install_skill.py context-engineering --project "D:\path\to\project" or package with python scripts/package_skill.py context-engineering.
Designs evidence-led content strategy, public-site information architecture, conversion and SEO structure, localization, visual communication requirements, and UX/UI handoffs for content-driven experiences.
Location: skills/content-strategy-architecture
Designs user-facing concepts, objects, capability groups, navigation, lists/details, findability, role visibility, localization, visual communication requirements, and UX handoffs for functional software applications.
Location: skills/product-information-architecture
Use Content Strategy & Architecture for public marketing/content surfaces and Product Information Architecture for authenticated or functional application surfaces. A mixed product may use both for separate surfaces without duplicating ownership.
Designs, analyses, documents, and reviews end-to-end application UX flows, including user needs, journeys, states, business rules, permissions, failures, recovery, accessibility, analytics, and implementation-ready acceptance criteria.
Location: skills/ux-flow-designer
Builds, refactors, and reviews React 19.x TypeScript with React Compiler assumed.
Location: skills/react-19
Designs, implements, reviews, and maintains lean risk-based automated test suites.
Location: skills/test-engineering
skills/<skill-name>/ is the catalog/source layout. Each skill is a portable, self-contained package with SKILL.md and its direct references.
The canonical installed layout is vendor-neutral:
Project: <project>/.agents/skills/<skill-name>/
Global: ~/.agents/skills/<skill-name>/
Use .agents/ as the single source of truth for installed skills. Do not maintain parallel .cursor/, .claude/, .codex/, or other vendor-specific copies of the same skill. Agent-specific tooling may point to, sync, or import the shared .agents/ tree when required by that tool.
This repository does not claim that every AI product natively auto-discovers .agents/; the environment using these skills is responsible for exposing the shared .agents/ installation to each agent.
Keep runtime references relative to the skill directory so the package can be moved, synchronized, zipped, or uploaded without repository-external dependencies.
From this repository root:
python scripts/install_skill.py software-architecture-advisor --project "D:\path\to\project"Installs to:
D:\path\to\project\.agents\skills\software-architecture-advisor\
python scripts/install_skill.py software-architecture-advisor --globalInstalls to:
~/.agents/skills/software-architecture-advisor/
The installer replaces only the selected skill directory and leaves other installed skills untouched.
For consumers that accept uploaded skill packages, create a neutral archive:
python scripts/package_skill.py software-architecture-advisorOutput:
dist/software-architecture-advisor.zip
The package keeps SKILL.md at the ZIP root, validates required frontmatter and relative references, and excludes source-only installation guidance.
For ChatGPT accounts/workspaces that expose uploaded Skills, the same neutral ZIP can be uploaded through Plugins → Skills → Create → Upload from your computer. ChatGPT upload is one consumer path; it is not the canonical repository installation model.
See skills/software-architecture-advisor/INSTALL.md for details.
Generated archives under dist/ are local build artifacts and are not committed.
skills/
software-architecture-advisor/
SKILL.md
PROFILE.md
PROJECT-CONTEXT.md
DECISION-FRAMEWORK.md
OUTPUT-TEMPLATE.md
INSTALL.md
doc-strategy-engineer/
context-engineering/
content-strategy-architecture/
product-information-architecture/
ux-flow-designer/
react-19/
test-engineering/
scripts/
install_skill.py
package_skill.py
docs/
improvements/
Each skill keeps its entry instructions concise and places detailed guidance in direct reference files or subdirectories owned by that skill.