cure-llm-ledger

Product Engineering Skills

The complete skill library that Cure Consulting Group uses to build apps, platforms, and products. These skills encode our standards, frameworks, and processes — so every project ships with the same level of rigor.

Now available as a Claude Code Plugin — install once, get auto-updates across all projects.

How It's Organized

ProductEngineeringSkills/
├── .claude-plugin/           # Plugin manifest
│   └── plugin.json
├── skills/{domain}/          # 80 skills, organized by domain (engineering/platform/product/business/marketing/security/legal)
│   ├── sdlc/
│   ├── android-feature-scaffold/
│   ├── incident-response/     # NEW
│   ├── accessibility-audit/   # NEW
│   ├── performance-review/    # NEW
│   ├── database-architect/    # NEW
│   ├── infrastructure-scaffold/ # NEW
│   ├── project-bootstrap/
│   ├── e2e-testing/
│   ├── test-accounts/
│   ├── uat/
│   ├── compliance-architect/
│   ├── data-migration/
│   ├── feature-flags/
│   ├── release-management/
│   ├── observability/
│   ├── client-handoff/
│   ├── llmops/
│   ├── disaster-recovery/
│   ├── dora-metrics/
│   ├── design-system/
│   ├── client-communication/
│   ├── i18n/
│   ├── notification-architect/
│   ├── offline-first/
│   ├── chaos-engineering/
│   ├── edge-computing/
│   ├── finops/
│   ├── micro-frontends/
│   ├── growth-engineering/
│   ├── green-software/
│   ├── proposal-generator/
│   ├── api-gateway/
│   ├── ... (75 total — see docs/OVERVIEW.md for full inventory)
│   └── legal-doc-scaffold/
├── agents/                   # 35 custom subagent definitions
├── personas/                 # 4 cross-domain engagement archetypes
│   ├── code-reviewer.md      # Security + quality review agent
│   ├── project-bootstrapper.md  # New project setup agent
│   ├── test-runner.md        # Execute test suites, report coverage
│   ├── pr-reviewer.md        # Automated PR diff review
│   ├── refactor-assistant.md # Safe refactoring with test validation
│   ├── ci-debugger.md        # Diagnose failed CI/CD runs
│   ├── release-coordinator.md # Version bump, changelog, deploy validation
│   ├── doc-generator.md      # API docs, ADRs, changelogs from code
│   ├── codebase-explainer.md # Onboarding — explain architecture, trace flows
│   ├── migration-validator.md # Database migration safety checks
│   ├── deployment-validator.md # Pre-deployment checklist validation
│   ├── dependency-auditor.md # Vulnerability and outdated package audit
│   ├── api-validator.md      # OpenAPI spec and contract validation
│   ├── product-analyst.md    # Feature adoption, analytics instrumentation
│   ├── ux-researcher.md      # Usability analysis, friction mapping
│   ├── roadmap-strategist.md # RICE scoring, dependency mapping, roadmaps
│   ├── competitive-intel.md  # Feature matrices, positioning, moat analysis
│   ├── content-strategist.md # Editorial calendars, SEO, content briefs
│   ├── campaign-analyst.md   # Attribution, funnel analysis, channel ROI
│   ├── brand-guardian.md     # Voice/tone, visual identity, microcopy audit
│   ├── growth-analyst.md     # Activation, retention, viral mechanics
│   ├── financial-analyst.md  # Revenue forecasts, unit economics, scenarios
│   ├── market-intelligence.md # TAM/SAM/SOM, trends, market timing
│   ├── investor-relations.md # Board updates, KPIs, fundraising narratives
│   ├── contract-reviewer.md  # SOW/contract risk, terms, IP review
│   ├── data-analyst.md       # Schema exploration, queries, data quality
│   ├── metrics-dashboard.md  # KPI definitions, SLOs, dashboard wireframes
│   ├── ab-test-analyst.md    # Experiment design, statistical analysis
│   ├── qa-engineer.md         # Test planning, edge cases, regression, quality gates
│   ├── accessibility-checker.md # WCAG 2.2 automated compliance
│   └── firebase-security-auditor.md # Firestore rules and Functions audit
├── hooks/                    # Multi-layer automated enforcement
│   └── hooks.json            # Command + Prompt hooks (9 event types)
├── rules/                    # 11 path-specific coding standards
│   ├── android.md             # Loads for *.kt files
│   ├── ios.md                 # Loads for *.swift files
│   ├── web.md                 # Loads for *.ts/*.tsx files
│   ├── firebase.md            # Loads for functions/**
│   ├── python.md              # Loads for *.py files
│   ├── go.md                  # Loads for *.go files
│   ├── rust.md                # Loads for *.rs files
│   ├── sql.md                 # Loads for *.sql, migrations/**
│   ├── docker.md              # Loads for Dockerfile, *.dockerfile
│   ├── terraform.md           # Loads for *.tf, *.tfvars
│   └── cicd.md                # Loads for .github/workflows/**
├── output-styles/            # 9 custom output formatting styles
│   ├── prd/                   # Product docs (PRDs, GTM, research)
│   ├── code-generation/       # Code scaffolds and implementations
│   ├── financial-analysis/    # Cost models, SaaS metrics
│   ├── audit-report/          # Audits, reviews, compliance
│   ├── api-specification/     # OpenAPI specs, endpoint docs
│   ├── architecture-decision/ # ADRs, RFCs, trade-off matrices
│   ├── runbook/               # Incident runbooks, DR procedures
│   ├── test-plan/             # Test plans, coverage reports
│   └── monitoring-alert/      # Alert definitions, thresholds
├── .mcp.json                 # MCP server configs (GitHub, Sentry, Firestore, PostgreSQL)
├── .lsp.json                 # LSP server configs (TypeScript, Python/Pyright)
├── marketplace.json          # Plugin marketplace manifest
├── settings.json             # Default permission rules
├── claude-commands/           # Legacy format (backwards compat, 64 files)
├── gemini skills/             # Google Gemini skills (.skill ZIP)
├── CLAUDE.md                  # Project instructions (Claude)
├── GEMINI.md                  # Project instructions (Gemini CLI)
├── AGENT-GUIDE.md             # How to structure prompts for agents & skills
├── setup.sh                  # Setup script for Antigravity & other projects
└── README.md

Installation

Via GitHub Package (Recommended)

Install the plugin as an npm package from GitHub Packages. This is the easiest way to keep all your projects up to date.

1. Authenticate with GitHub Packages (one-time setup):

# Create a Personal Access Token (PAT) with read:packages scope at
# https://github.com/settings/tokens, then:
npm login --scope=@cure-consulting-group --registry=https://npm.pkg.github.com

Or add to your project's .npmrc:

@cure-consulting-group:registry=https://npm.pkg.github.com
//npm.pkg.github.com/:_authToken=${GITHUB_TOKEN}

2. Install in your project:

npm install @cure-consulting-group/product-engineering-skills

The postinstall script automatically:

• Symlinks the package to ~/.claude/plugins/ProductEngineeringSkills
• Registers the plugin in ~/.claude/settings.json

All 80 skills, 39 agents, 4 personas, hooks, rules, and output styles are immediately available.

3. Enable auto-updates with Dependabot (recommended):

Add .github/dependabot.yml to your project (or run setup.sh which does this automatically):

version: 2
updates:
  - package-ecosystem: "npm"
    directory: "/"
    schedule:
      interval: "daily"
    allow:
      - dependency-name: "@cure-consulting-group/product-engineering-skills"
    labels:
      - "dependencies"
      - "skills-update"
    commit-message:
      prefix: "chore"
      include: "scope"

Dependabot will open a PR in your project whenever a new version is published. Merge it and every agent on that project gets the updated skills.

4. Manual update:

npm update @cure-consulting-group/product-engineering-skills

As a Claude Code Plugin (Manual)

# Load the plugin during a session
claude --plugin-dir /path/to/ProductEngineeringSkills

# Or for development/testing
claude --plugin-dir ./ProductEngineeringSkills

Once loaded, all skills are available as namespaced commands:

/cure-product-engineering:sdlc
/cure-product-engineering:feature-audit
/cure-product-engineering:android-feature-scaffold
/cure-product-engineering:incident-response
/cure-product-engineering:accessibility-audit

Hooks, agents, rules, output styles, and MCP servers are all included automatically.

Setup Script (Antigravity & Other Projects)

The fastest way to onboard any project:

# From the target project directory
/path/to/ProductEngineeringSkills/setup.sh

# Or specify the project path
/path/to/ProductEngineeringSkills/setup.sh /path/to/antigravity-app

# Install globally for ALL projects
/path/to/ProductEngineeringSkills/setup.sh --global

# Legacy mode (just copy skills, no hooks/agents)
/path/to/ProductEngineeringSkills/setup.sh --legacy

The setup script will:

1. Clone/update the plugin to ~/.claude/plugins/
2. Detect your project type (Android/iOS/Web/Firebase)
3. Install only the relevant path-specific rules
4. Create a project CLAUDE.md with skill references
5. Add .claude/settings.local.json to .gitignore

Via Marketplace

# Add the Cure Consulting marketplace
claude marketplace add https://github.com/Cure-Consulting-Group/ProductEngineeringSkills/marketplace.json

# Install the plugin
claude plugin install cure-product-engineering

Legacy Method (Copy Commands)

Copy the claude-commands/ files into your project's .claude/commands/ directory:

cp claude-commands/*.md /path/to/your/project/.claude/commands/

Then use them as slash commands:

/sdlc — Generate SDLC artifacts
/android-feature-scaffold — Scaffold an Android feature module
/feature-audit — Audit a completed feature

Using with Google Gemini

Import the .skill files from gemini skills/ into your Gemini workspace. Each .skill file is a ZIP archive containing:

• SKILL.md — The main skill definition
• references/ — Supporting documents and templates

Skill Inventory (64 Skills)

Product & Strategy (7)

Engineering & Architecture (18)

Quality & Security (11)

Operations & Delivery (12)

Business & Finance (7)

Portfolio Management (2) — NEW

Consulting Operations (3)

Platform Design (4)

Recurring Automation (Loops & Routines)

The library ships a standard maintenance loop (self-provisioned to .claude/loop.md on first session start — run bare /loop to use it), Recurring Mode sections in the goal-shaped skills (finops, burn-rate-tracker, investor-reporting, security-review, and others), and copy-paste cloud-routine recipes in docs/AUTOMATION.md. Library upkeep cadence: docs/MAINTENANCE.md. Mechanism selection and unattended-run guardrails: /cure-product-engineering:engagement-automation.

Hooks (Multi-Layer Automated Enforcement)

The plugin ships command and prompt hooks across 9 event types: SessionStart, PreCompact, PostCompact, ConfigChange, PostToolUseFailure, UserPromptSubmit, PreToolUse, Stop, SubagentStop. Highlights: a Stop-hook quality gate (blocks "done" without verification), a PreToolUse static security guard on skill/agent/persona files, and a ConfigChange audit trigger when skill files change mid-session.

New in v4.0: Hooks now suggest and auto-trigger agents based on context. Every code edit, test run, deployment, and PR action recommends the most relevant agent(s).

Command Hooks (Deterministic)

Prompt Hooks (LLM-Validated)

Agent Hooks (Multi-Turn Verification)

MCP Server Integrations

Pre-configured MCP servers in .mcp.json:

LSP Server Integrations

Pre-configured LSP servers in .lsp.json:

Output Styles

Custom output formatting for different artifact types:

Custom Agents (30)

Engineering Agents (14)