Agent-Skills.md

Agent Skills: Feature Implementation

[Implementation] Use when the user asks to implement a new feature, enhancement, add functionality, build something new, or create new capabilities. Triggers on keywords like "implement", "add feature", "build", "create new", "develop", "enhancement".

UncategorizedID: duc01226/easyplatform/feature

Repository

duc01226/easyplatform
duc01226License: MIT
31

Install this agent skill to your local

pnpm dlx add-skill https://github.com/duc01226/EasyPlatform/tree/HEAD/.claude/skills/feature

Skill Files

Browse the full folder contents for feature.

Download Skill

Loading file tree…

.claude/skills/feature/SKILL.md

Skill Metadata

Name
feature
Description
"[Implementation] Use when the user asks to implement a new feature, enhancement, add functionality, build something new, or create new capabilities. Triggers on keywords like "implement", "add feature", "build", "create new", "develop", "enhancement"."

Feature Implementation

Expert full-stack .NET + Angular implementation skill with structured investigation, planning, and layered execution.

Summary

Goal: Implement new features end-to-end with structured investigation, layered execution (Domain > Persistence > Application > API > Frontend), and verification.

| Step | Action | Key Notes | |------|--------|-----------| | 1 | Design Intent | State WHAT+WHY, risks, guiding principle in 3 sentences | | 2 | Investigate | Decompose requirements, search codebase for related patterns | | 3 | Knowledge graph | Build analysis in .ai/workspace/analysis/{feature}.md | | 4 | Plan | Layered implementation plan (backend bottom-up, frontend service-first) | | 5 | Approval gate | Present plan -- DO NOT proceed without user approval | | 6 | Execute | Implement layer-by-layer following approved plan | | 7 | Verify | Type checks, tests, integration validation |

Key Principles:

  • Evidence-based: verify assumptions with grep/search before implementing
  • Platform-first: use EasyPlatform patterns; CQRS Command+Result+Handler in ONE file
  • Logic in LOWEST layer: Entity > Service > Component

Mode Selection

| Mode | Flag | Behavior | | ------------------------- | -------------- | ------------------------------------------------------------------------------------ | | Interactive (default) | (none) | Approval gates at plan + execution; user feedback loop | | Autonomous | --autonomous | Structured headless workflow; approval gate at plan only, then execute to completion |

Workflow Overview

  1. Design Intent -- State WHAT+WHY, risks, guiding principle in 3 sentences
  2. Investigate -- Decompose requirements, search codebase for related entities/patterns/boundaries
  3. Knowledge Graph -- Build structured analysis in .ai/workspace/analysis/{feature}.md
  4. Plan -- Generate layered implementation plan (Domain > Persistence > Application > API > Frontend)
  5. Approval Gate -- Present plan for user approval; do NOT proceed without it
  6. Execute -- Implement layer-by-layer following approved plan
  7. Verify -- Type checks, tests, integration validation

⚠️ MUST READ .claude/skills/feature/references/implementation-workflow.md for detailed phase instructions.

⚠️ MUST READ Before Implementation

IMPORTANT: You MUST read these files before starting. Do NOT skip.

  • ⚠️ MUST READ .claude/skills/shared/anti-hallucination-protocol.md — validation checkpoints
  • ⚠️ MUST READ .claude/skills/shared/knowledge-graph-template.md — per-file analysis structure
  • ⚠️ MUST READ CLAUDE.md — codebase instructions and platform patterns

Implementation Order

Backend (bottom-up)

  1. Domain Layer -- Entity, expressions, value objects
  2. Persistence Layer -- Configuration, migration
  3. Application Layer -- DTOs, commands, queries, event handlers
  4. API Layer -- Controller endpoints

Frontend (service-first)

  1. API Service -- PlatformApiService extension
  2. Store -- PlatformVmStore state management
  3. Components -- AppBaseVmStoreComponent / AppBaseFormComponent
  4. Routing -- Route definitions and guards

Approval Gate Format

## Implementation Plan Complete - Approval Required

### Summary
[Brief description]

### Files to Create
1. `path/to/file` - [purpose]

### Files to Modify
1. `path/to/file:line` - [change description]

### Implementation Order
1. [Step 1] ... N. [Step N]

### Risks & Considerations
- [Risk 1]

**Awaiting approval to proceed with implementation.**

Execution Safeguards

  • Verify file exists before modification
  • Read current content before editing
  • Check conflicts with existing code
  • Validate changes against platform patterns
  • If any step fails: HALT, report failure, return to approval gate

Post-Implementation

  • Verify against all requirements
  • Document under ## Success Validation heading
  • Summarize changes in changelog.md

⚠️ MUST READ .claude/skills/feature/references/validation-checklist.md for verification items. ⚠️ MUST READ .claude/skills/feature/references/ep-file-locations.md for path reference and related skills.

Coding Guidelines

  • Evidence-based: verify assumptions with grep/search
  • Platform-first: use established EasyPlatform patterns
  • Cross-service: use event bus, never direct DB access
  • CQRS: Command + Result + Handler in ONE file
  • Logic in LOWEST layer: Entity > Service > Component

IMPORTANT Task Planning Notes

  • Always plan and break many small todo tasks
  • Always add a final review todo task to review the works done at the end to find any fix or enhancement needed