Incorporating Everything Claude Code (ECC) into the NovaTrek Continuous Architecture Platform¶

Date: 2026-03-17 Status: In Progress Author: Solution Architecture Team

Decisions Made¶

Demo Rollback Methodology¶

The demo creates or modifies files during live demonstration. To reset cleanly:

# Before demo: save a checkpoint
./scripts/demo-reset.sh save

# Run the demo (invoke prompts, show outputs)

# After demo: reset to checkpoint
./scripts/demo-reset.sh reset

# Check what would be reset (without resetting)
./scripts/demo-reset.sh status

Executive Summary¶

This plan describes how to incorporate the Everything Claude Code (ECC) repository — a mature, community-driven AI agent harness performance system — into the NovaTrek Continuous Architecture Platform's GitHub Copilot AI workflow. ECC provides 108 skills, 25 specialized agents, 57 commands, automated hooks, MCP configurations, and language-specific rules that can significantly enhance the quality, consistency, and capability of AI-assisted architecture work.

The integration targets GitHub Copilot Agent Mode running in VS Code, adapting ECC's Claude Code-oriented patterns to the Copilot instruction and skill system (.github/copilot-instructions.md, .instructions.md, .prompt.md files).

Table of Contents¶

1. Current State Assessment
2. ECC Asset Inventory — Relevant to NovaTrek
3. Integration Strategy
4. Phase 1 — Foundation (Immediate)
5. Phase 2 — Architecture Skills (Short-Term)
6. Phase 3 — Advanced Workflows (Medium-Term)
7. Phase 4 — Continuous Learning (Longer-Term)
8. Skill Translation Guide — Claude Code to Copilot
9. Demonstration Plan
10. Risk Assessment
11. File Manifest — What Gets Created or Modified

1. Current State Assessment¶

What We Have Today¶

What ECC Adds¶

2. ECC Asset Inventory — Relevant to NovaTrek¶

Tier 1 — Directly Applicable (Incorporate Now)¶

These skills map directly to activities our Solution Architect AI performs daily.

Tier 2 — Valuable Enhancements (Short-Term)¶

Tier 3 — Future Value (Medium-Term)¶

3. Integration Strategy¶

Approach: Adaptation, Not Direct Copy¶

ECC skills are designed for Claude Code's agent harness (subagent delegation, hooks, slash commands). GitHub Copilot uses a different mechanism:

Key Principles¶

1. Preserve NovaTrek domain specificity — ECC's generic patterns are adapted to our adventure tourism microservices domain, not used as-is
2. Augment, do not replace — Our existing copilot-instructions.md is comprehensive and domain-specific; ECC adds workflow patterns and quality gates, not domain knowledge
3. Gradual rollout — Incorporate Tier 1 first, validate in real scenarios, then expand
4. Source attribution — All adapted content references the ECC origin for license compliance (MIT)

4. Phase 1 — Foundation (Immediate)¶

4.1 Create Architecture Skill Instructions¶

Create folder-scoped .instructions.md files that Copilot loads automatically when working in specific directories.

4.1.1 Solution Design Instructions¶

File: architecture/solutions/.instructions.md

Adapts from: agents/architect.md, agents/planner.md, skills/search-first/

Content: - Architecture review process (current state analysis, requirements gathering, design proposal, trade-off analysis) - Prior-art discovery workflow (search-first pattern from ECC) - MADR decision-making framework (already in copilot-instructions, reinforced here) - Solution decomposition (from planner agent)

4.1.2 API Spec Design Instructions¶

File: architecture/specs/.instructions.md

Adapts from: skills/api-design/

Content: - REST API design patterns (resource naming, status codes, pagination, filtering) - OpenAPI spec quality checklist (schema completeness, nullable annotations, enum validation) - Backward compatibility rules for API changes - Error response standardization

4.1.3 Security Review Instructions¶

File: architecture/.instructions.md

Adapts from: skills/security-review/, rules/common/security.md

Content: - OWASP Top 10 checklist adapted for NovaTrek services - PII handling rules for guest profiles and waivers - Authentication and authorization patterns (svc-guest-profiles as identity source) - Input validation at service boundaries

4.2 Create Reusable Prompt Files¶

4.2.1 Deep Research Prompt¶

File: .github/prompts/deep-research.prompt.md

Adapts from: skills/deep-research/

Content: - Multi-source research methodology - Source attribution requirements - Evidence-based findings with workspace file references - Synthesis and structured output format

4.2.2 Architecture Review Prompt¶

File: .github/prompts/architecture-review.prompt.md

Adapts from: agents/architect.md, skills/security-review/

Content: - Current state analysis checklist - Trade-off documentation template - ISO 25010 quality attribute assessment - Anti-pattern detection (from copilot-instructions, enhanced with ECC patterns)

4.2.3 Investigation Prompt¶

File: .github/prompts/investigation.prompt.md

Adapts from: contexts/research.md, skills/search-first/

Content: - Research mode activation (understand before acting) - Mock tool usage sequence (JIRA first, Elastic before GitLab) - Evidence gathering and citation methodology - Hypothesis formation and verification

4.3 Augment Copilot Instructions¶

Add an "AI Workflow Patterns" section to .github/copilot-instructions.md that incorporates:

• Search-first principle (from skills/search-first/): Always search for existing solutions before writing new code or creating new designs
• Research mode (from contexts/research.md): When investigating tickets, switch to exploration mode — read widely, form hypotheses, verify with evidence
• Security-first (from rules/common/security.md): Check for security implications in every solution design touching auth, PII, or cross-service data flows
• Context management (from skills/strategic-compact/): For long architecture sessions, compact context at logical breakpoints

5. Phase 2 — Architecture Skills (Short-Term)¶

5.1 Java and Spring Boot Analysis Skills¶

File: source-code/.instructions.md

Adapts from: skills/java-coding-standards/, skills/springboot-patterns/, skills/jpa-patterns/

Content: - Java coding standards for NovaTrek source code analysis - Spring Boot patterns (dependency injection, entity/repository, transactions) - JPA/Hibernate anti-pattern detection (N+1 queries, entity replacement, missing @Version) - Spring Security patterns for authentication and authorization

5.2 Database Design Skills¶

File: architecture/metadata/.instructions.md

Adapts from: skills/postgres-patterns/, skills/database-migrations/

Content: - PostgreSQL schema design best practices - Database migration strategy (no-downtime, rollback strategies) - Index optimization patterns - Data store documentation standards for NovaTrek services

5.3 Docker and Local Development¶

File: docker-compose.yml folder .instructions.md

Adapts from: skills/docker-patterns/

Content: - Docker Compose patterns for multi-service orchestration - Container security best practices - Volume management for persistent data - Network configuration for service isolation

5.4 Enhanced Security Review¶

File: .github/prompts/security-review.prompt.md

Adapts from: skills/security-review/, skills/springboot-security/

Content: - Full OWASP Top 10 checklist - Spring Boot-specific security patterns - NovaTrek PII handling (guest profiles, waivers, payment data) - Cross-service authentication patterns - Severity rating methodology (Critical/High/Medium/Low)

6. Phase 3 — Advanced Workflows (Medium-Term)¶

6.1 Verification Loop for Solution Designs¶

Adapts from: skills/verification-loop/, skills/springboot-verification/

Create a verification prompt that runs a quality gate on solution designs before they are merged:

• All affected services identified with specific API/schema changes
• MADR ADRs created for cross-boundary decisions
• Impact assessments address WHAT changes (not HOW)
• User stories written from user perspective
• ISO 25010 quality attributes assessed
• Backward compatibility addressed
• Prior-art referenced

6.2 Continuous Learning Integration¶

Adapts from: skills/continuous-learning/, skills/continuous-learning-v2/

Implement pattern extraction from architecture sessions:

• After each solution design session, extract reusable patterns
• Store patterns in architecture/reminders/ (existing location)
• Build instinct library: atomic observations that evolve into architectural heuristics
• Examples: "When svc-check-in is involved, always check Pattern 3 safety fallback" or "Cross-domain calls require event-driven integration"

6.3 Multi-Agent Workflow Patterns¶

Adapts from: skills/autonomous-loops/, skills/enterprise-agent-ops/

Document structured multi-agent workflows for complex scenarios:

• Scenario 1 — Full Ticket Investigation: Research agent explores ticket, Architect agent designs solution, Security reviewer agent validates
• Scenario 2 — Impact Analysis: One agent per affected service, results consolidated by architect agent
• Scenario 3 — ADR Authoring: Research agent gathers options, Architect agent evaluates trade-offs, Documentation agent formats MADR

6.4 Context Management Strategy¶

Adapts from: skills/strategic-compact/

Implement context checkpointing for long architecture sessions:

• Save progress to session memory at logical breakpoints
• Compact context when approaching limits
• Resume from checkpoints with full architectural context
• Use architecture/reminders/ for persistent cross-session notes

7. Phase 4 — Continuous Learning (Longer-Term)¶

7.1 Architecture Instinct Library¶

Build a NovaTrek-specific instinct library based on ECC's continuous learning patterns:

7.2 Eval-Driven Architecture Quality¶

Adapts from: skills/eval-harness/

Create evaluation criteria for AI-generated architecture artifacts:

• Solution Design Completeness: Does it cover all required sections?
• ADR Quality: Are at least 2 genuine options considered? Are consequences documented?
• Impact Assessment Accuracy: Are affected services correctly identified? Are API changes specific?
• Prior-Art Reference: Is the capability changelog consulted? Are related solutions referenced?

7.3 Cost-Aware Model Routing for Architecture Tasks¶

Adapts from: skills/cost-aware-llm-pipeline/

Optimize model selection for different architecture tasks:

8. Skill Translation Guide — Claude Code to Copilot¶

How to Adapt an ECC Skill for Copilot¶

Step 1: Read the ECC SKILL.md file - Extract the "When to Activate" triggers - Extract the "Core Workflow" steps - Extract the "Reference Guide" lookup table

Step 2: Determine the Copilot target format

Step 3: Adapt the content - Replace Claude Code-specific references (subagent delegation, hooks, slash commands) - Add NovaTrek domain context (service names, data ownership, safety rules) - Reference workspace files instead of generic examples - Include file paths to relevant specs, metadata, and source code

Step 4: Test the skill - Run a scenario that would trigger the skill - Verify the AI follows the skill's workflow - Check that domain-specific context is applied correctly

Example Translation¶

ECC Source (skills/api-design/SKILL.md):

## When to Activate
- Designing new API endpoints
- Reviewing existing API contracts

Copilot Adaptation (architecture/specs/.instructions.md):

## When Working in This Directory
When creating or modifying OpenAPI specs in this directory:
1. Follow REST API design patterns (resource naming, status codes, pagination)
2. Validate schema completeness: all fields need types, descriptions, and nullable annotations
3. Check backward compatibility: new required fields break existing consumers
4. Cross-reference with data ownership boundaries in architecture/metadata/data-ownership.yaml
5. Verify enum values against NovaTrek domain values (adventure categories, check-in patterns)

9. Demonstration Plan¶

Demo 1 — API Design Skill in Action¶

Objective: Show how the adapted API design skill improves OpenAPI spec quality.

Setup: 1. Create architecture/specs/.instructions.md with adapted API design patterns 2. Open an existing OpenAPI spec (e.g., architecture/specs/svc-check-in.yaml)

Scenario: Ask Copilot to "Add a new endpoint for group check-in to svc-check-in"

Expected Behavior (with ECC skill): - Follows resource naming conventions (POST /check-ins/group) - Includes proper status codes (201 Created, 400 Bad Request, 409 Conflict) - Adds pagination for list endpoints - Documents nullable fields - Cross-references with svc-guest-profiles for identity resolution - Applies Pattern 3 safety fallback rule

Comparison (without skill): - May use inconsistent naming - May omit error responses - May not consider NovaTrek domain rules

Demo 2 — Research-First Investigation¶

Objective: Show how the research-first workflow improves ticket investigation quality.

Setup: 1. Create .github/prompts/investigation.prompt.md with research mode 2. Use a ticket that requires cross-service investigation

Scenario: "Investigate NTK-10005 using the investigation workflow"

Expected Behavior (with ECC skill): - Activates research mode (understand before acting) - Runs JIRA mock tool first for ticket context - Runs Elastic mock tool for production logs - Runs GitLab mock tool for related MRs - Forms hypothesis based on evidence - Documents findings with specific file paths and line numbers - Produces structured investigation report

Demo 3 — Security Review Enhancement¶

Objective: Show how the security review skill catches issues in solution designs.

Setup: 1. Create .github/prompts/security-review.prompt.md 2. Use a solution design that involves PII or authentication

Scenario: "Run a security review on a selected solution design"

Expected Behavior (with ECC skill): - Applies OWASP Top 10 checklist - Checks PII handling (guest profiles, waiver data) - Validates authentication flow through svc-guest-profiles - Identifies missing input validation at service boundaries - Rates findings by severity - Produces structured security report

Demo 4 — Continuous Learning Capture¶

Objective: Show how architectural insights are captured and reused across sessions.

Setup: 1. Create a learning capture prompt 2. Complete a solution design session

Scenario: After completing a solution design, run "Capture architectural learnings from this session"

Expected Behavior: - Extracts reusable patterns (e.g., "Group check-in requires orchestrator pattern") - Stores in architecture/reminders/ with timestamps - On next session, loads relevant reminders as context - Patterns evolve into architectural instincts over time

Demo 5 — Before/After Comparison¶

Objective: Side-by-side comparison of AI output quality with and without ECC skills.

Method: 1. Run the same architecture task twice: - Run A: Default Copilot with existing copilot-instructions.md only - Run B: Copilot with ECC-adapted skills loaded 2. Compare output on these dimensions:

10. Risk Assessment¶

11. File Manifest — What Gets Created or Modified¶

New Files to Create¶

Files to Modify¶

Reference Files (Read-Only)¶

The everything-claude-code/ directory remains as a read-only reference. Skills are adapted, not symlinked.

Appendix A — ECC Skill Format Reference¶

SKILL.md Structure (Claude Code)¶

---
name: skill-name
description: When and why to use this skill
origin: ECC
---

# Skill Name
## When to Activate
## Core Workflow
## Reference Guide
## Rules and Conventions

Copilot .instructions.md Structure¶

---
applyTo: "**/*.yaml"  # optional: restrict to file types
---

# Context for AI
## When Working in This Directory
## Quality Checklist
## Domain Rules

Copilot .prompt.md Structure¶

---
mode: "agent"
description: "Description shown in prompt picker"
---

# Workflow Name
## Steps
## Output Format
## Validation

Appendix B — Full ECC Skills Catalog (108 Skills)¶

See the complete catalog in everything-claude-code/README.md or the detailed exploration notes. The skills are grouped into:

• Architecture and Design (11 skills) — api-design, backend-patterns, frontend-patterns, docker-patterns, android-clean-architecture, swiftui-patterns, kotlin-ktor-patterns, kotlin-exposed-patterns, kotlin-coroutines-flows, swift-actor-persistence, swift-concurrency-6-2
• Security (5 skills) — security-review, security-scan, django-security, laravel-security, springboot-security
• Documentation and Research (8 skills) — deep-research, article-writing, market-research, investor-materials, investor-outreach, documentation-lookup, regex-vs-llm-structured-text, iterative-retrieval
• Testing and Quality (16 skills) — tdd-workflow, springboot-tdd, e2e-testing, django-tdd, django-verification, laravel-verification, springboot-verification, cpp-testing, rust-testing, golang-testing, kotlin-testing, python-testing, verification-loop, eval-harness, test-coverage, quality-nonconformance
• DevOps and Deployment (4 skills) — deployment-patterns, database-migrations, bun-runtime, nextjs-turbopack
• Code Quality and Patterns (12 skills) — coding-standards, java-coding-standards, python-patterns, golang-patterns, rust-patterns, cpp-coding-standards, perl-patterns, laravel-patterns, django-patterns, springboot-patterns, plankton-code-quality, strategic-compact
• AI/Agent Engineering (12 skills) — agentic-engineering, autonomous-loops, continuous-agent-loop, continuous-learning, continuous-learning-v2, enterprise-agent-ops, skill-stocktake, team-builder, ralphinho-rfc-pipeline, nanoclaw-repl, agent-harness-construction, configure-ecc
• Domain-Specific and Business (13 skills) — Various industry verticals
• Foundation Models and APIs (4 skills) — foundation-models-on-device, claude-api, cost-aware-llm-pipeline, prompt-optimizer
• Infrastructure and Search (4 skills) — clickhouse-io, postgres-patterns, exa-search, search-first
• Other (19 skills) — Various specialized skills

Appendix C — ECC Agents Catalog (25 Agents)¶