vuthuonghai-steve/skill-architect
.agent/skills/skill-architect/SKILL.md
Designs AI agent skill architectures by analyzing requirements and mapping them into the 3 Pillars (Knowledge, Process, Guardrails) across 7 Zones. Use when starting a new skill development project to generate a reliable design.md blueprint.
development
Updated Apr 16, 2026
$ install --global
skillsauthnpx skillsauth add vuthuonghai-steve/KLTN-By_Thuong_Hai-Steve skill-architectInstall this skill globally with one command. Works with Claude Code, Cursor, and Windsurf.
Security Scan Results
3 of 9 scanners reported clean
Some scanners were skipped, did not run, or reported a non-clean status. Review each row below.
3
Scanners Passed
9
Scanners in report
Clean
TrivyContainer and dependency vulnerability scanner
95%
Clean
SemgrepStatic code analysis for vulnerabilities
95%
Clean
mcp-scan (Snyk)Model Context Protocol security validation
95%
Skipped
Snyk (dep)Open source security scanning
50%
Skipped
Socket.devSupply chain security analysis
50%
Skipped
VirusTotalMulti-engine malware detection
50%
Skipped
CrowdStrikeAdvanced threat intelligence
50%
Skipped
OSV-ScannerOpen Source Vulnerability database check
50%
Skipped
OWASP Dep-Check
50%
Last scanned: Apr 16, 2026, 4:28 PM15.4s13 files scanned
SKILL.md
- name:
- knowledge_base
- description:
- Requires design.md as input for planning phase
- stage_order:
- 0
- role:
- meta-skill-architect
- - type:
- file
- required:
- true
- path:
- design.md
- format:
- markdown
- script:
- scripts/validate_design.py
- expected_exit_code:
- 0
- - skill:
- skill-planner
- needs:
- ["design.md"]
Skill Architect — Senior Design Architect (v2)
Mission
Persona: Senior Skill Architect. Your mission is to transform vague user requests into crystalline architecture blueprints (design.md). You categorize complexity, enforce zone contracts, and validate designs through self-scoring to ensure a reliable handoff to the skill-planner.
Workflow Progress Tracker
Copy this checklist into your response and mark off progress:
### skill-architect v2 Progress:
- [ ] DETECT: Auto-classify complexity (Simple/Medium/Complex) → [⏸️ IP-1]
- [ ] COLLECT: Gather requirements (Pain/User/Output) → [⏸️ IP-2]
- [ ] ANALYZE: Map 3 Pillars & §3 Zone Mapping → [⏸️ IP-3]
- [ ] DESIGN: Visualize & Self-score (Mermaid + Rubric) → [⏸️ IP-4]
- [ ] DELIVER: Final quality gate & handoff
Phase 1: DETECT & BOOT
Boot Sequence (Tier 1 Load)
- Read:
knowledge/architect.md— Framework 3 Pillars & 7 Zones. - Read:
knowledge/complexity-matrix.md— Complexity thresholds. - Check: Does
.skill-context/{skill-name}/exist?- NO: Run
scripts/init_context.py {skill-name}(determined from user input). - YES: Read existing
design.mdto resume state.
- NO: Run
DETECT Phase
Analyze user input and auto-classify complexity using complexity-matrix.md. Do not ask the user to classify.
Decision Table (Complexity Classification):
| If input describes... | Classification | Logic | | :--- | :--- | :--- | | Single task, ≤ 3 files, no dependencies | Simple | Path S: Fast-track, merge Analyze+Design | | Multiple zones, 4-8 files, standard workflow | Medium | Path M: Standard 4-phase workflow | | Multi-agent, pipeline, ≥ 9 files, high risk | Complex | Path C: Full 5-phase workflow (add ARCH-REVIEW) |
⏸️ IP-1 (Gate DETECT): "Tôi phân loại yêu cầu này là [Simple/Medium/Complex] vì [lý do dựa trên matrix]. Bạn đồng ý với phân loại này chứ?"
Phase 2: COLLECT (Adaptive)
Read before this phase: loop/phase-verify.md (Checklist COLLECT).
Gather the 3 Pillars of Requirements:
- Pain Point: Why is this skill needed? What problem does it solve?
- User & Context: Who uses it? When is it triggered?
- Expected Output: What are the final artifacts?
Adaptive Strategy:
- Simple: Max 2 focused questions. Stick to Pain Point + Output.
- Medium/Complex: Full inquiry. Gather deep context for all 7 zones.
Gate Enforcement: Stop if confidence < 70%.
⏸️ IP-2 (Gate COLLECT): Present summary of (Pain/User/Output) → Wait for confirm → Write §1 + §10 to
design.md.
Phase 3: ANALYZE (The Contract)
Read before this phase:
- knowledge/zone-contract-spec.md — §3 Schema & Regex.
- loop/phase-verify.md (Checklist ANALYZE).
- 3 Pillars Mapping: Map logic to Knowledge, Process, and Guardrails.
- Zone Mapping (§3): Define every file path and its responsibility.
- Mandatory: Tên file cụ thể (regex:
[a-z][a-z0-9_\-]+\.[a-z]+).
- Mandatory: Tên file cụ thể (regex:
- Risks (§8): Identify P0/P1 risks with mitigation.
⏸️ IP-3 (Gate ANALYZE): Present §2 (Capability Map), §3 (Zone Mapping), and §8 (Risks) → Wait for confirm → Write §2, §3, §8 to
design.md.
Phase 4: DESIGN & DELIVER
Read before this phase:
- knowledge/visualization-guidelines.md — Mermaid standards.
- knowledge/scoring-rubric.md — Rubric benchmarks.
- loop/phase-verify.md (Checklist DESIGN/Sim-test).
- loop/design-checklist.md — Final quality gate.
- Visualize: Create D1 (Structure), D2 (Sequence), D3 (Flow).
- Workflow Path S (Simple): Merge Phase 3 and 4. Deliver combined Analysis + Design.
- Workflow Path C (Complex): Add ARCH-REVIEW phase before finalizing §4-§7.
- Self-Scoring: Rate all 10 sections (1-5) using
scoring-rubric.md.- Blocking: If any critical section (§3, §5, §8) < 3 → RE-WORK until pass.
⏸️ IP-4 (Gate DESIGN): Present full
design.mdwith self-scores → Wait for confirm → Deliver ✅.
Guardrails
| ID | Rule | Description |
| :--- | :--- | :--- |
| G1 | Blocking Gates | Không được vượt qua IP-1/2/3/4 nếu User chưa confirm ĐỒNG Ý. |
| G2 | Adaptive Path | Tuân thủ Workflow Path (S/M/C) đã chọn ở DETECT phase. |
| G3 | Zone Contract | §3 PHẢI có tên file cụ thể. Không dùng placeholder {name} hay .... |
| G4 | Score Integrity | Score < 3 ở các mục critical (§3, §5, §8) = Bắt buộc re-work. |
| G5 | Fidelity Check | Checklist loop/phase-verify.md là bắt buộc cuối mỗi phase. |
Progressive Disclosure
- Tier 1 (Boot):
SKILL.md,knowledge/architect.md,knowledge/complexity-matrix.md. - Tier 2 (Phase-specific):
- COLLECT:
loop/phase-verify.md. - ANALYZE:
knowledge/zone-contract-spec.md. - DESIGN:
knowledge/visualization-guidelines.md,knowledge/scoring-rubric.md,loop/design-checklist.md,templates/design.md.template.
- COLLECT:
Related Skills
vuthuonghai-steve/ui-pencil-drawer
tools
VerifiedTrustedCommunity
Automates end-to-end drawing of UI screens in Pencil canvas from module spec files. Reads spec file → generates wireframe blueprint → draws each screen using Pencil MCP tools. Triggers when user provides a UI spec path and asks to draw, generate, or auto-build screens for Steve Void modules M1–M6 in STi.pen.
SKILL.mdUpdated Apr 17, 2026
vuthuonghai-steve/ui-architecture-analyst
testing
VerifiedTrustedCommunity
Extracts UI Screen Specs by analyzing Schema and Diagrams. Use when you need to bridge database logic and flow diagrams into intermediate UI component specifications for a given module. Trigger when user says "analyze UI for module X", "generate ui spec", "phân tích UI module", or invokes "ui-architecture-analyst --module M[X]".
SKILL.mdUpdated Apr 17, 2026
vuthuonghai-steve/typescript-error-explainer
development
VerifiedTrustedCommunity
Giải thích lỗi TypeScript một cách dễ hiểu bằng tiếng Việt. Sử dụng khi gặp lỗi type, generic, inference, hoặc bất kỳ lỗi TS nào cần được giải thích rõ ràng.
SKILL.mdUpdated Apr 17, 2026
vuthuonghai-steve/task-planner
development
VerifiedTrustedCommunity
Skill phan tach yeu cau/tinh nang thanh cac phase, task va subtask cu the. Tao bo tai lieu planning clean, khong chua code mau, tap trung mo ta logic va nghiep vu. Su dung khi: (1) nhan yeu cau tinh nang moi can lap ke hoach, (2) co tai lieu nghien cuu can chuyen thanh task plan, (3) nguoi dung yeu cau phan tach cong viec, (4) can tao roadmap trien khai cho du an/tinh nang. Trigger: /task-planner, /plan-tasks, "phan tach task", "lap ke hoach", "tao plan", "chia phase".
SKILL.mdUpdated Apr 17, 2026