dependency first architect

Stop building software based on what is most exciting and start building based on what is most stable. This skill transforms vague build requests into rigorous, dependency-ordered architectures that prioritize the substrate before the surface. It eliminates the rework caused by unvalidated assumptions and ensures that hard-to-reverse decisions are settled through explicit validation gates rather than silent guesses.

What it does

• Dependency sequencing orders tasks by structural, runtime, and economic blocks to prevent out-of-order development.
• Tradeoff gating classifies decisions by reversibility (R1-R3) and forces explicit validation for high-risk choices.
• Walking skeleton design defines the thinnest end-to-end production path to verify deployments, logging, and rollbacks on day zero.
• Blast radius management schedules work to ensure high-impact rework happens first while exposure to users happens last.
• Integrated hardening threads security, observability, and resilience into every phase rather than treating them as final steps.

How it works

1. System classification determines the system type, dominant constraints, and the most critical failure modes.
2. Tradeoff analysis identifies R3 decisions like data models or API contracts that require early validation.
3. Phase construction generates a sequenced plan starting with Phase 0 (the skeleton) and moving through dependency-cleared stages.
4. Validation mapping creates a formal gate table with hypotheses, thresholds, and evidence requirements for every milestone.

Frameworks & tools

This skill is framework-agnostic and applies to greenfield development, brownfield migrations, and complex AI/agentic systems involving RAG, LLM orchestration, and human-in-the-loop gating.

Why this beats prompting it yourself

Generic prompts usually produce a simple chronological checklist that ignores architectural dependencies and risk. This skill enforces a "blast radius" logic that identifies where a project is most likely to fail early, preventing weeks of rework on leaf nodes when the root substrate is flawed.

Use cases

• Infrastructure migrations for moving legacy systems to cloud-native stacks with minimal downtime and verified rollbacks.
• AI agent development to sequence cost and latency budgeting before building orchestration layers.
• Greenfield product launches where structural data models and auth boundaries must be validated before the UI is built.
• Regulatory compliance builds that require security controls to be verified before any real data enters the environment.

Known limitations

This skill focuses on architectural sequencing rather than generating implementation code. It requires the user to provide specific budget constraints or accept assumed values that must be validated later.