Skip to content

README.md

Latest commit

Β 

History

History
187 lines (158 loc) Β· 7.11 KB

README.md

File metadata and controls

187 lines (158 loc) Β· 7.11 KB
Raw

Dev Kit Agent Skill

Full-stack Development Kit theo chuαΊ©n Agent Skills β€” tΖ°Ζ‘ng thΓ­ch Claude Code, Cursor, VS Code, Kiro, Gemini CLI, vΓ  40+ agentic clients khΓ‘c.

CΓ i Δ‘αΊ·t

Claude Code

CΓ‘ch 1 β€” Global (dΓΉng được ở mọi project):

git clone git@github.com:SoftwareOneHN/dev-kit.git ~/.claude/skills/dev-kit

CΓ‘ch 2 β€” Project-level (chỉ Γ‘p dα»₯ng cho 1 project):

git clone git@github.com:SoftwareOneHN/dev-kit.git .claude/skills/dev-kit

CΓ‘ch 3 β€” Symlink (dev workflow, dα»… pull updates):

git clone git@github.com:SoftwareOneHN/dev-kit.git ~/Projects/dev-kit
ln -s ~/Projects/dev-kit ~/.claude/skills/dev-kit

Global vs Project-level: ~/.claude/skills/ lΓ  global β€” skill tα»± Δ‘α»™ng available trong mọi project. .claude/skills/ (trong repo) lΓ  project-level β€” chỉ hoαΊ‘t Δ‘α»™ng khi bαΊ‘n Δ‘ang ở repo Δ‘Γ³.

Verify: ls ~/.claude/skills/dev-kit/SKILL.md β€” thαΊ₯y file lΓ  OK.

Cursor / VS Code

Copy thΖ° mα»₯c dev-kit/ vΓ o .cursor/skills/ hoαΊ·c .vscode/skills/.

Kiro

Copy thΖ° mα»₯c dev-kit/ vΓ o .kiro/skills/ hoαΊ·c cΓ i tα»« Power registry.

BαΊ₯t kα»³ Agent Skills client nΓ o

Đặt thΖ° mα»₯c dev-kit/ vΓ o vα»‹ trΓ­ skills mΓ  client hα»— trợ.

CαΊ₯u trΓΊc

dev-kit/
β”œβ”€β”€ SKILL.md              # Required: metadata + core instructions
β”œβ”€β”€ README.md             # File nΓ y
β”œβ”€β”€ LICENSE               # MIT
└── references/           # Detailed docs (loaded on-demand)
    β”œβ”€β”€ _shared/
    β”‚   └── workflow-artifacts.md
    β”œβ”€β”€ languages/
    β”‚   β”œβ”€β”€ java/
    β”‚   β”‚   β”œβ”€β”€ core-rules.md
    β”‚   β”‚   β”œβ”€β”€ spring-boot.md
    β”‚   β”‚   β”œβ”€β”€ clean-architecture.md
    β”‚   β”‚   └── testing.md
    β”‚   └── typescript/
    β”‚       β”œβ”€β”€ core-rules.md
    β”‚       β”œβ”€β”€ react-nextjs.md
    β”‚       β”œβ”€β”€ vue-nuxt.md
    β”‚       β”œβ”€β”€ api-patterns.md
    β”‚       └── forms.md
    β”œβ”€β”€ brainstorm/
    β”‚   └── SKILL.md
    β”œβ”€β”€ spec/
    β”‚   └── SKILL.md
    β”œβ”€β”€ plan/
    β”‚   β”œβ”€β”€ SKILL.md
    β”‚   β”œβ”€β”€ archive-workflow.md
    β”‚   β”œβ”€β”€ codebase-understanding.md
    β”‚   β”œβ”€β”€ output-standards.md
    β”‚   β”œβ”€β”€ plan-organization.md
    β”‚   β”œβ”€β”€ red-team-personas.md
    β”‚   β”œβ”€β”€ red-team-workflow.md
    β”‚   β”œβ”€β”€ research-phase.md
    β”‚   β”œβ”€β”€ solution-design.md
    β”‚   β”œβ”€β”€ task-management.md
    β”‚   β”œβ”€β”€ validate-question-framework.md
    β”‚   β”œβ”€β”€ validate-workflow.md
    β”‚   └── workflow-modes.md
    β”œβ”€β”€ cook/
    β”‚   β”œβ”€β”€ SKILL.md
    β”‚   β”œβ”€β”€ intent-detection.md
    β”‚   β”œβ”€β”€ review-cycle.md
    β”‚   β”œβ”€β”€ subagent-patterns.md
    β”‚   └── workflow-steps.md
    β”œβ”€β”€ fix/
    β”‚   β”œβ”€β”€ SKILL.md
    β”‚   β”œβ”€β”€ complexity-assessment.md
    β”‚   β”œβ”€β”€ diagnosis-protocol.md
    β”‚   β”œβ”€β”€ mode-selection.md
    β”‚   β”œβ”€β”€ parallel-exploration.md
    β”‚   β”œβ”€β”€ prevention-gate.md
    β”‚   β”œβ”€β”€ review-cycle.md
    β”‚   β”œβ”€β”€ skill-activation-matrix.md
    β”‚   β”œβ”€β”€ task-orchestration.md
    β”‚   β”œβ”€β”€ workflow-ci.md
    β”‚   β”œβ”€β”€ workflow-deep.md
    β”‚   β”œβ”€β”€ workflow-logs.md
    β”‚   β”œβ”€β”€ workflow-quick.md
    β”‚   β”œβ”€β”€ workflow-standard.md
    β”‚   β”œβ”€β”€ workflow-test.md
    β”‚   β”œβ”€β”€ workflow-types.md
    β”‚   └── workflow-ui.md
    β”œβ”€β”€ debug/
    β”‚   β”œβ”€β”€ SKILL.md
    β”‚   β”œβ”€β”€ defense-in-depth.md
    β”‚   β”œβ”€β”€ frontend-verification.md
    β”‚   β”œβ”€β”€ investigation-methodology.md
    β”‚   β”œβ”€β”€ log-and-ci-analysis.md
    β”‚   β”œβ”€β”€ performance-diagnostics.md
    β”‚   β”œβ”€β”€ reporting-standards.md
    β”‚   β”œβ”€β”€ root-cause-tracing.md
    β”‚   β”œβ”€β”€ systematic-debugging.md
    β”‚   β”œβ”€β”€ task-management-debugging.md
    β”‚   └── verification.md
    β”œβ”€β”€ test/
    β”‚   β”œβ”€β”€ SKILL.md
    β”‚   β”œβ”€β”€ report-format.md
    β”‚   β”œβ”€β”€ test-execution-workflow.md
    β”‚   └── ui-testing-workflow.md
    β”œβ”€β”€ scout/
    β”‚   β”œβ”€β”€ SKILL.md
    β”‚   β”œβ”€β”€ external-scouting.md
    β”‚   β”œβ”€β”€ internal-scouting.md
    β”‚   └── task-management-scouting.md
    └── code-review/
        β”œβ”€β”€ SKILL.md
        β”œβ”€β”€ adversarial-review.md
        β”œβ”€β”€ checklist-workflow.md
        β”œβ”€β”€ checklists/
        β”‚   β”œβ”€β”€ api.md
        β”‚   β”œβ”€β”€ base.md
        β”‚   └── web-app.md
        β”œβ”€β”€ code-review-reception.md
        β”œβ”€β”€ codebase-scan-workflow.md
        β”œβ”€β”€ edge-case-scouting.md
        β”œβ”€β”€ input-mode-resolution.md
        β”œβ”€β”€ parallel-review-workflow.md
        β”œβ”€β”€ requesting-code-review.md
        β”œβ”€β”€ spec-compliance-review.md
        β”œβ”€β”€ task-management-reviews.md
        └── verification-before-completion.md

Progressive Disclosure

Theo chuαΊ©n Agent Skills, skill hoαΊ‘t Δ‘α»™ng theo 3 giai Δ‘oαΊ‘n:

  1. Discovery (~100 tokens): Chỉ load name + description tα»« frontmatter
  2. Activation (~4000 tokens): Load toΓ n bα»™ SKILL.md body khi task match
  3. Execution (on-demand): Load reference files cα»₯ thể khi cαΊ§n chi tiαΊΏt step

Sα»­ dα»₯ng

Sau khi cΓ i, nΓ³i tα»± nhiΓͺn:

BαΊ‘n nΓ³i Agent lΓ m
"Brainstorm cΓ‘ch implement auth" Scout β†’ exact requirements β†’ trade-off analysis β†’ design doc
"Plan feature user profile" Research β†’ codebase analysis β†’ solution design β†’ phase plan
"Implement story login tα»« PRD" Scout β†’ plan β†’ code β†’ test β†’ review β†’ finalize
"Fix lα»—i 500 ở endpoint /users" Scout β†’ diagnose root cause β†’ fix β†’ verify+prevent
"Debug tαΊ‘i sao test flaky" Systematic debugging β†’ root cause trace β†’ report
"Run tests cho module auth" Execute test suite β†’ coverage β†’ QA report
"Review PR #42" Fetch diff β†’ spec compliance β†’ quality β†’ red-team
"Scout codebase cho payment module" Parallel agents β†’ file discovery β†’ structured report

Key Features

  • Lifecycle-driven: brainstorm β†’ plan β†’ implement β†’ test β†’ review β†’ ship
  • Scout-first: always scan codebase before asking questions or making changes
  • Evidence-based: no guessing, no "probably" β€” verify before claiming
  • Adversarial: red-team review on every code change
  • Side-effect-free: mandatory verification that changes don't break existing behavior
  • Multi-mode: interactive, auto, fast, parallel workflows for different needs
  • Language-agnostic: works with any language, framework, or tech stack
  • Task-managed: native Claude Tasks for tracking multi-step workflows

License

MIT