Policy-as-Code Framework for Claude Code Enterprise
A four-layer configuration framework that codifies engineering standards and SDLC best practices into Claude Code artifacts — skills, agents, MCP servers, hooks, rules, and managed settings.
Built for tech services teams delivering data platforms and data products across multiple clients on AWS, GCP, Azure, Databricks, and Snowflake.
┌─────────────────────────────────────────────────────────┐
│ Layer 4: CLIENT / ENGAGEMENT │
│ Per-project .claude/ directory │
│ Client EA patterns, Jira/Confluence config, SOPs │
├─────────────────────────────────────────────────────────┤
│ Layer 3: PLATFORM │
│ Installable modules (opt-in per engagement) │
│ AWS / Azure / GCP / Databricks / Snowflake │
├─────────────────────────────────────────────────────────┤
│ Layer 2: DOMAIN │
│ Data engineering, analytics, ML/DS, full-stack data │
│ Data quality, pipeline patterns, schema design │
├─────────────────────────────────────────────────────────┤
│ Layer 1: CORE (universal, immutable per client) │
│ Coding standards, SDLC orchestrators, testing, │
│ formatting, security, governance │
└─────────────────────────────────────────────────────────┘
Layer 1 (Core) never changes per client — it represents our engineering standards.
Layers 2-4 are configured per engagement via the /client-onboard skill.
git clone https://github.com/Diz312/cc-framework.git
cd cc-framework
./scripts/install.sh# In Claude Code, run:
/client-onboardThis walks you through a structured Q&A to configure the framework for the client's environment (EA patterns, cloud platform, security, Jira/Confluence, data governance).
/discovery # Collect requirements (green-field or brown-field)
/design # Create solution architecture
/build # Implement the solution
/test # Comprehensive testing
/deploy # Ship it| Skill | Phase | Purpose |
|---|---|---|
/client-onboard |
Pre-project | Configure framework for client environment |
/discovery |
Discovery | Requirements gathering, current-state analysis |
/design |
Design | Solution architecture, constrained by EA |
/build |
Build | Implementation with standards enforcement |
/test |
Test | Unit, data quality, schema, security testing |
/deploy |
Deploy | PR creation, CI/CD, ticket updates |
| Skill | Purpose |
|---|---|
/format-and-lint |
Python code formatting (black, ruff, mypy) |
/test-runner |
pytest with coverage reporting |
/whitepaper |
Render markdown to consulting-grade PDF (WeasyPrint, custom typography) |
| Agent | Purpose |
|---|---|
solution-architect |
Propose architecture options (EA-aware, research-backed) |
requirements-collector |
Standardized requirements gathering |
brownfield-analyzer |
Current-state analysis for existing systems |
framework-verifier |
Verify framework APIs before coding |
test-writer |
Write comprehensive pytest suites |
schema-designer |
Database schema design (Domain: data-engineering) |
api-integrator |
FastAPI REST API builder (Domain: full-stack-data) |
| Rule | Scope |
|---|---|
security.md |
Security patterns, secrets handling, OWASP |
git-workflow.md |
Branching, commits, PR standards |
code-review.md |
Review checklist and standards |
| Platform | Status |
|---|---|
| GCP | MVP |
| AWS | Planned |
| Azure | Planned |
| Databricks | Planned |
| Snowflake | Planned |
| Domain | Status |
|---|---|
| Data Engineering | MVP |
| Analytics/BI | MVP |
| ML/Data Science | Planned |
| Full-Stack Data | Planned |
cc-framework/
├── core/ # Layer 1: Universal (immutable per client)
│ ├── CLAUDE.md # Global engineering standards
│ ├── CODING_STANDARDS.md # Authoritative coding standards
│ ├── settings.json # Default permissions
│ ├── skills/ # Phase orchestrators + utility skills
│ ├── agents/ # Universal sub-agents
│ ├── rules/ # Always-on constraints
│ └── tools/ # CLI tools backing skills
├── domain/ # Layer 2: Domain-specific
│ ├── data-engineering/ # Data engineering standards
│ ├── analytics/ # Analytics/BI standards
│ ├── ml-ds/ # ML/Data Science standards
│ └── full-stack-data/ # Full-stack data app standards
├── platform/ # Layer 3: Cloud platform modules
│ ├── aws/
│ ├── azure/
│ ├── gcp/
│ ├── databricks/
│ └── snowflake/
├── client/ # Layer 4: Engagement templates
│ ├── templates/ # Scaffolding templates
│ ├── examples/ # Example client setups
│ └── onboarding/ # Onboarding guides
├── scripts/ # Installation and management
├── sops/ # Standard Operating Procedures
└── docs/ # Documentation
├── getting-started/
├── architecture/
├── sdlc/
├── playbooks/
├── reference/
└── whitepaper/
- Layer 1 is immutable — Core engineering standards never change per client
- Platform-native first — Maximize client's cloud platform before external tools
- Agentic + HIL — Agents propose, humans approve
- Green-field / brown-field aware — Discovery adapts to project context
- EA-constrained design — Architecture honors client enterprise patterns
- Standardized artifacts — Each phase produces artifacts feeding the next
- Auditable — All engagement configuration is version-controlled
This framework's design is informed by:
- DORA 2025: "AI is an amplifier, not a fix" — governance must precede AI tooling
- NIST AI RMF: Embed governance in workflow, not bolted on
- ISO 42001: International standard for AI management systems
- GitHub Copilot Enterprise: Policy cascade pattern (Enterprise > Org > User)
- Deloitte 2026: Only 1 in 5 companies have mature AI agent governance
- KPMG Data Product Lifecycle: Standardized phases for data product delivery
See docs/whitepaper/ for the full research-backed analysis.
See CONTRIBUTING.md for how to add skills, agents, platform modules, and domain extensions.
Internal use. See LICENSE for details.