ai-tools-workflow

name	ai-tools-workflow
description	Creates structured, complete project outputs with any coding AI tool (Codex, Copilot, Cursor, Claude, or ChatGPT). Activates when writing prompts for AI tools, generating README/project guides, planning implementation tasks, or asking for complete documentation with architecture, roles, objectives, setup, testing, and FAQs.

AI Tools Workflow

When to Apply

Activate this skill when:

• You want high-quality results from any coding AI tool.
• You need a complete README or project guide.
• You are writing reusable prompts for Codex/Copilot/Cursor/Claude/ChatGPT.
• You want outputs that include architecture, conventions, setup, testing, and maintenance notes.

Skill Composition For UI Tasks

When a task includes interactive UI and styling, apply both skills together in the same response workflow:

• livewire-development: component structure, state, events, wire directives, and Livewire tests.
• tailwindcss-development: layout, spacing, responsive behavior, and theme token usage.

Use this rule for tasks like modals, forms, tables, filters, profile pages, dashboard widgets, and validation error UI.

Prompt line you can reuse:

Use livewire-development for behavior and tailwindcss-development for styling. Keep color usage token-based from the shared app palette.

Activation Flow (No-Code First)

Quick invoke phrase:

Start with memory gate: Review -> Solution -> Plan -> Confirmation (no code yet) -> Execution.

Use this sequence before any implementation work:

1. Review
   • Read .github/memory.md and compare it to the current request.
   • Decide if the request is connected to the last session.
2. Solution
   • State Connected or Not connected in 1-2 lines with a reason.
3. Plan
   • Provide the task plan and boundaries.
4. Confirmation (no code yet)
   • Ask for explicit approval before editing files or running implementation commands.
5. Execution
   • Implement only after confirmation.

Connected Statement Template

Use this when the request is connected to the last session:

Continuing from last session: this task extends [previous task/topic], so I will preserve [invariant/constraint] while implementing [current request].

Not Connected Statement Template

Use this when the request is not connected:

This request is not directly connected to the last session in .github/memory.md, so I will proceed as a fresh task while keeping existing project conventions intact.

Universal Prompt Template

Use this template in any AI tool.

You are helping document and maintain this software project.

Task:
Create a complete DILMS for this repository.

Context:
- Project name: DILMS
- Stack: LARAVEL, LIVEWIRE, ALPINE, TAILWIND, PHPUNIT
- Main users/roles: ADMIN,SUPERVISOR, STUDENT
- Business objective: 
Improve attendance accuracy and reduce missed or inconsistent logs.
Enforce policy-driven attendance workflows without manual checking.
Give admins and supervisors a complete view of OJT participation.
Make reporting faster, clearer, and easier to audit.
Preserve a trustworthy record of important actions.

Output requirements:
1. Write in clear Markdown with professional headings.
2. Explain purpose, scope, and target users.
3. Describe tech stack and why each tool exists.
4. Define roles, permissions, and responsibilities.
5. Explain architecture and component boundaries.
6. Show end-to-end flow (request -> logic -> persistence -> reporting/UI).
7. Include setup, run, test, and build commands with short explanations.
8. Add environment/config section with required vs optional variables.
9. Include testing strategy (feature/unit and critical scenarios).
10. Document invariants, pitfalls, and troubleshooting.
11. Add FAQ and roadmap/future improvements.
12. If details are missing, mark assumptions clearly.

Quality bar:
- Do not be generic; tie explanations to this project structure.
- Prefer concise but complete sections.
- Keep consistency with existing conventions and architecture.
- Provide diagrams (Mermaid) where they add clarity.

Deliverables:
- Final Markdown content only.
- Then a short checklist showing coverage of all required sections.

README-Focused Prompt (Drop-In)

Create a production-ready README.md for this repository.

Must include:
- Project purpose and problem statement
- Roles and responsibilities
- Business objectives and success criteria
- Full tech stack table
- Architecture explanation and Mermaid diagrams
- Data flow and module boundaries
- Setup/run/test/build commands
- Environment variables table (required vs optional)
- Testing strategy and key scenarios
- Known pitfalls and troubleshooting
- FAQ (at least 10 items)
- Contributing guidelines and PR definition of done
- Glossary and roadmap

Constraints:
- Align with existing code conventions and folder structure.
- Prefer practical details over generic framework explanations.
- Mark unknowns as Assumption.

API Documentation Prompt

Generate API documentation for this repository.
Include:
- Endpoint definitions with request/response examples
- Authentication and authorization requirements
- Error codes and handling
- Rate limits and best practices
- Example API flows for common use cases

Output Review Checklist

Use this checklist after every AI-generated document:

1. Is the project purpose explicit and measurable?
2. Are all user roles and permissions defined?
3. Are commands accurate for this repo?
4. Is architecture mapped to real folders/files?
5. Are invariants and business rules documented?
6. Are testing expectations clear and realistic?
7. Are pitfalls/troubleshooting specific to this project?
8. Are assumptions labeled clearly?
9. Is language concise and non-redundant?
10. Is the document actionable for a new contributor?

Common Mistakes to Avoid

• Asking for "a README" without constraints.
• Accepting framework-generic content unrelated to the repository.
• Missing setup/test commands and environment variables.
• Omitting architecture boundaries and invariants.
• Failing to include contributor workflow and done criteria.

Pro Tip

Run AI in two passes:

1. Pass 1: Generate a full draft with all required sections.
2. Pass 2: Ask the AI to verify every section against the checklist and fix gaps.