navigate-compass
Adaptive Engineering Scoping & Architecture Planning
- ✓ When scoping a greenfield software project from scratch.
- ✓ When planning a major redesign, architectural overhaul, or database migration.
- ✓ When decomposing large, ambiguous engineering initiatives into actionable implementation roadmaps.
- ✓ When defining clear technical architecture and dependency relationships before writing code.
Trigger: Activated when the user requests project planning, architectural design, roadmap scoping, or feature breakdowns.
"Plan a real-time collaborative whiteboarding system from scratch."
Adaptive Socratic Interview & Tri-Artifact Graph Synthesis
Comprehensive planning context with adaptive multi-round conversation history.
Stores output artifacts in docs/plan/ (plan.md, dependency-graph.md, and plan-graph.json).
Reference Specifications (`references/`)
Multi-round questioning strategy covering architecture, data modeling, concurrency, and security.
Standardized 11-section markdown template ensuring comprehensive, audit-ready planning.
Mermaid syntax rules and JSON adjacency list schema for automated downstream ingestion.
Executive Overview
Premature implementation is the most common reason software projects suffer from scope creep, architectural rework, and technical debt. navigate-compass solves this by forcing a rigorous, consultant-grade scoping process before a single line of application code is written.
Acting as a System Architect and Technical Consultant, navigate-compass runs an adaptive Socratic interview tailored to four distinct project scenarios:
- Greenfield Development: Defining tech stack, high-level topology, MVP boundaries, and database schemas.
- System Redesign: Analyzing pain points in legacy code and designing backwards-compatible transitions.
- Refactoring Initiatives: Isolating code smells, defining bounded contexts, and establishing characterization test suites.
- Feature Additions: Ensuring new capabilities integrate seamlessly with existing service contracts.
The Tri-Artifact Output System
Upon concluding the adaptive interview, navigate-compass synthesizes three distinct artifacts into docs/plan/:
1. docs/plan/plan.md (The Master Specification)
A comprehensive, 11-section engineering blueprint covering:
- Executive Summary & Problem Statement
- MVP Scope vs. Full Vision Roadmap
- System Architecture & Component Diagram
- Tech Stack Justification & Trade-off Analysis
- Data Models, Schemas & Entity Relationships
- API & Event Contracts
- Error Handling & Resilience Strategy
- Testing & Quality Assurance Strategy
- Prioritized Task Breakdown Table
- Security & Compliance Audit Checklist
2. docs/plan/dependency-graph.md (Mermaid Visualization)
A visual flowchart displaying task dependencies, execution paths, and critical blockers:
[Task 1: DB Migrations] --> [Task 2: Repository Layer] --> [Task 4: Controller]
↓
[Task 3: Unit Tests]
3. docs/plan/plan-graph.json (Graph-Prompting Adjacency List)
A machine-readable JSON structure designed specifically for downstream automated agent processing (split-plan and execute-issues):
{
"project": "Geospatial Discovery Engine",
"nodes": [
{ "id": "task-001", "title": "Setup MongoDB 2dsphere Index", "deps": [] },
{ "id": "task-002", "title": "Implement Snaps Gateway", "deps": ["task-001"] }
]
}
Adaptive Interview Methodology
Rather than asking static, rigid questionnaires, navigate-compass dynamically adapts based on developer responses:
- If high concurrency is specified, it immediately drills into connection pooling, WebSocket gateways, and message broker backpressure.
- If relational data is specified, it probes normalization, foreign keys, cascade behaviors, and indexing strategies.
- If external API integrations are required, it challenges failure models, circuit breakers, and idempotency guarantees.