Agent Skills Kit

A portable agent skills toolkit, deployable on Claude Code, Cursor AI, Continue.dev, GitHub Copilot, and OpenAI Codex

Quick Start (30 seconds)

Option 1: npm (Recommended)

npx @hyper_dev/skills

Or install globally:

npm install -g @hyper_dev/skills
hyper-dev-skills

Option 2: Local Installation

./scripts/install.sh

Select platform → confirm → done!

What's Included

26 Agent Skills

Skill	Description	Category
bidding-orchestrator	End-to-end hybrid bidding package orchestration	Orchestration
bid-evidence-hub	MCP + internet evidence aggregation for bidding	Analysis
bid-solution-designer	Baseline vs optimized solution design for proposals	Synthesis
bid-quality-gates	Proposal lifecycle quality gate evaluation	Analysis
bid-estimator	Best/base/worst PM and hybrid cost range estimation	Analysis
bid-staffing-planner	Phase-based staffing strategy generation	Analysis
bid-slide-factory	Bid deck brief and slide artifact generation	Documentation
deep-codebase-discovery	Orchestrator for full pipeline discovery	Orchestration
repo-recon	Analyzing the Repository Structure	Analysis
tech-build-audit	Audit tech stack & build system	Analysis
module-summary-report	Architecture & Module summary	Synthesis
reverse-doc-reconstruction	Create Reverse Documentation from code	Documentation
legacy-cpp-porting-guardrails	Port C++ code safely	Porting
cpp-java-migration-planner	Dependency-aware module/file/function migration order planning for C++ to Java	Porting
cpp-java-pre-porting	Pre-porting analysis and compat-layer design for C++ to Java migration	Porting
cpp-java-file-structure-porting	1:1 file skeleton migration from C++ to Java	Porting
cpp-java-function-porting	1:1 function conversion workflow from C++ to Java	Porting
cpp-java-porting-orchestrator	End-to-end orchestration of staged C++ to Java migration	Orchestration
bug-impact-analyzer	Bug Analysis & Impact Assessment	Bug Analysis
knows	Unified Git + MCP + memory evidence retrieval	Analysis
harness-builder	Build AI coding agent harness infrastructure (AGENTS.md, verification, lifecycle)	Infrastructure
wiki-generator	Generate structured wiki pages from code and docs (Index, Architecture, Module, API, Setup)	Documentation
hi-security	STRIDE + OWASP security audit with MCP-assisted analysis and iterative auto-fix	Security
hi-scenario	12-dimension edge case and scenario explorer before implementation	Analysis
hi-plan	Multi-mode technical implementation planning with red-team review	Planning
hi-predict	5-persona pre-analysis debate to catch issues before implementation	Analysis

👉 Detailed Usage Guide → — Comprehensive guide for each skill with examples, inputs, outputs, and use cases.

Multi-Platform Support

Platform	Installation	Usage
Claude Code	--agent claude-code	/skill deep-codebase-discovery
Cursor AI	--agent cursor --scope global	Ask normally, active via .cursorrules
Continue.dev	--agent continue	Ctrl+Shift+A → Select skill
GitHub Copilot	--agent copilot	Auto-select from .github/skills/*
OpenAI CodeX	--agent codex	Copy prompt to ChatGPT/CodeX

This will check:

• ✓ Claude Code: ~/.claude/skills/*/
• ✓ Cursor global: ~/.cursorrules
• ✓ Cursor local: .cursorrules
• ✓ Continue.dev: ~/.continue/skills/*.json
• ✓ Copilot: .github/skills/*/SKILL.md
• ✓ CodeX: .openai/codex-instructions.md
• ✓ Templates: template/ directory with 28 files

Project Structure

agent-skill/
├── bidding-orchestrator/           # Project bidding orchestrator
│   ├── SKILL.md
│   ├── contracts/
│   ├── examples/
│   ├── references/
│   └── scripts/
├── bid-evidence-hub/
├── bid-solution-designer/
├── bid-quality-gates/
├── bid-estimator/
├── bid-staffing-planner/
├── bid-slide-factory/
├── deep-codebase-discovery/        # Skill directories
│   ├── SKILL.md                     # Main skill definition
│   ├── references/                  # Supporting docs & playbooks
│   │   └── mcp-orchestration-playbook.md
│   └── scripts/                     # Skill-specific scripts (optional in source)
├── repo-recon/
├── tech-build-audit/
├── module-summary-report/
├── reverse-doc-reconstruction/
├── legacy-cpp-porting-guardrails/
├── cpp-java-migration-planner/
├── cpp-java-pre-porting/
├── cpp-java-file-structure-porting/
├── cpp-java-function-porting/
├── cpp-java-porting-orchestrator/
├── bug-impact-analyzer/
├── harness-builder/                 # AI agent harness engineering
│   ├── SKILL.md
│   ├── agents/openai.yaml
│   ├── references/                  # 6 pattern docs (memory, context, tool-registry...)
│   ├── templates/                   # 7 harness file templates
│   └── scripts/harness_init.py
├── wiki-generator/                  # Code/docs → structured wiki pages
│   ├── SKILL.md
│   ├── agents/openai.yaml
│   ├── references/wiki-page-template.md
│   ├── templates/                   # 5 wiki page templates (index, architecture, module, api, setup)
│   └── scripts/wiki_bootstrap.py
├── hi-security/                     # STRIDE + OWASP security audit
│   ├── SKILL.md
│   ├── agents/openai.yaml
│   ├── references/stride-owasp-checklist.md
│   └── scripts/secret_scan.py
├── hi-scenario/                     # 12-dimension edge case explorer
│   ├── SKILL.md
│   ├── agents/openai.yaml
│   └── references/dimension-checklist.md
├── hi-plan/                         # Multi-mode technical planning
│   ├── SKILL.md
│   ├── agents/openai.yaml
│   └── references/                  # 5 planning reference docs
├── hi-predict/                      # 5-persona pre-analysis debate
│   ├── SKILL.md
│   ├── agents/openai.yaml
│   └── references/persona-playbook.md
│
├── template/                        # standard templates (28 files)
│   ├── 00_requirements/            # Requirements templates
│   │   ├── tpl_requirements_spec.md
│   │   └── tpl_feature_list.md
│   ├── 01_usecase/                 # Use case templates
│   │   ├── tpl_usecase_list.md
│   │   ├── tpl_usecase_detail.md
│   │   └── tpl_usecase_metrics.md
│   ├── 02_detail_design/           # Design templates
│   │   ├── tpl_screen_design.md
│   │   ├── tpl_api_process_design.md
│   │   ├── tpl_openapi_spec.yaml
│   │   ├── tpl_table_design.md
│   │   ├── tpl_sql_design.md
│   │   └── tpl_batch_process_design.md
│   ├── 03_system_design/           # Architecture templates
│   ├── 04_testing/                 # Testing templates
│   ├── 05_operations/              # Ops templates
│   └── 06_project_mgmt/            # PM templates
│
├── scripts/                         # Installer & tools
│   ├── install.sh                  # Shell installer (traditional)
│   ├── install_agent_kit.py        # Python installer (traditional)
│   ├── install_agent_kit_with_templates.py  #  Enhanced installer
│   ├── convert_skill.py            # Format converter
│   ├── skill_parser.py             # Shared skill parser
│   ├── sync_manifest.py            # Manifest sync tool
│   ├── smoke_test.sh               # Smoke test suite
│   └── verify_install.sh           # Verification script
│
├── test/                             # Test artifacts
│   └── TEST_RESULTS.md              # Installation and smoke test results
├── verify/                           # Review and verification docs
│   ├── QUICK_START.md               # Quick start guide
│   ├── TEMPLATE_QUICKSTART.md       # Template quick start
│   ├── TEMPLATE_INSTALLATION_GUIDE.md
│   ├── TEMPLATE_INTEGRATION_GUIDE.md
│   ├── INSTALLER_README.md
│   ├── CROSS_PLATFORM_GUIDE.md
│   ├── SECURITY_REVIEW_REPORT.md
│   └── ...
├── MANIFEST.json                    # Skill metadata
└── README.md                        # This file

---

📖 Full Documentation

→ Complete Usage Guide with all 26 skills →

Topics covered:

• ✅ Installation methods (npm, shell, Python)
• ✅ Each skill: purpose, when to use, how to use, inputs, outputs
• ✅ MCP requirements & validation rules
• ✅ Skill chaining workflows
• ✅ Troubleshooting guide

---

📦 Installation Methods

Method 1: npm (Recommended)

# Quick install
npx @hyper_dev/skills

# Or install globally
npm install -g @hyper_dev/skills
hyper-dev-skills

Why npm?

• ✅ One-command installation
• ✅ Auto-detects your platforms (Claude Code, Cursor, Copilot, etc.)
• ✅ Self-updates
• ✅ No git clone required

Method 2: Shell Script (Local)

git clone https://github.com/hyper-dev/agent-skill.git
cd agent-skill
./scripts/install.sh

Method 3: Python Installer (Advanced)

python scripts/install_agent_kit_with_templates.py \
  --agent claude-code \
  --include-templates

Method 4: Manual (Platform-Specific)

Claude Code:

cp -r {bidding-orchestrator,deep-codebase-discovery,...} ~/.claude/skills/

GitHub Copilot:

cp -r {bidding-orchestrator,deep-codebase-discovery,...} .github/skills/

Cursor:

# Global scope
cp -r .cursorrules ~/.cursorrules

# Local scope (per-project)
cp -r .cursorrules .cursorrules

---

Contributing

1. Fork the repository
2. Create feature branch
3. Add/modify skills in */SKILL.md
4. Sync manifest: python scripts/sync_manifest.py
5. Run smoke tests: ./scripts/smoke_test.sh
6. Optional verification: ./scripts/verify_install.sh
7. Submit PR

Adding New Skills

1. Create directory: your-skill/SKILL.md
2. Add metadata in YAML frontmatter
3. Add references/ if needed
4. Review quality using template/06_project_mgmt/tpl_skill_review_checklist_mcp_first.md
5. Run python scripts/sync_manifest.py
6. Test installer

Resources

Core Features

• MCP Integration: Skills support mind_mcp and graph_mcp
• Context Control: Strategy for large codebases
• Template System: 28 standard templates across 6 categories
• Security: Comprehensive redaction patterns (see SECURITY_REVIEW_REPORT.md)

Templates

• Standard Templates: template/ directory with requirements, use cases, design, testing, operations, PM templates
• Template Usage: How skills use templates (see TEMPLATE_INTEGRATION_GUIDE.md)
• Template Installation: Installing skills with templates (see TEMPLATE_QUICKSTART.md)
• Skill Review Checklist: template/06_project_mgmt/tpl_skill_review_checklist_mcp_first.md

Tools

• Converter: Convert between skill formats
• Manifest Sync: Keep MANIFEST.json in sync with skills
• Smoke Tests: Validate skill quality

Troubleshooting

Permission denied

chmod +x ./scripts/install.sh

Claude Code can not detect skills

# check installation
ls ~/.claude/skills/*/

# Restart Claude Code CLI

Cursor can not detect skills

# check .cursorrules
cat ~/.cursorrules  # global
cat .cursorrules    # local

# Restart Cursor editor

📄 License

MIT License - feel free to use and modify

---