high-level-design

Generates committed HLD documents with C4 diagrams, ADRs, threat models, and capacity plans.

Updated May 21, 2026
One-click install
npx skills add https://github.com/CagesThrottleUs/private-ai-harness --skill high-level-design-cagesthrottleus
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: high-level-design
Source: https://github.com/CagesThrottleUs/private-ai-harness/tree/main/skills/high-level-design
Command: npx skills add https://github.com/CagesThrottleUs/private-ai-harness --skill high-level-design-cagesthrottleus

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Architecture decisions made in conversation get lost, and features with new services, data models, or integrations often proceed without a documented design, leading to misaligned implementation and unreviewed security or capacity risks. ## Core Features & Use Cases - Structured HLD Generation: Produces a committed HLD document with C4 Context and Container diagrams in Mermaid, technology selection records, STRIDE threat model, failure mode analysis, and capacity planning. - ADR Authoring: Writes one immutable Architecture Decision Record per significant decision in Nygard/MADR format, committed to wiki/architecture/ with global monotonic numbering. - Gated Review Workflow: Runs an hld-reviewer agent before human review and hard-blocks transition to implementation planning until explicit human approval. - Use Case: After a feature spec passes quality review, use this Skill to design a new payment service — it asks targeted architecture questions, diagrams the containers and trust boundaries, documents why PostgreSQL was chosen over alternatives, and produces ADRs before any code is planned. ## Quick Start Ask the AI to run the high-level-design skill for the approved spec to produce the HLD document, C4 diagrams, and ADRs before writing implementation plans.

Frequently Asked Questions about high-level-design

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I create a high-level design document for a new feature?

Provide a spec that has passed the spec-quality-gate, then the skill asks targeted architecture questions, presents a component decomposition for approval, and generates the HLD with C4 diagrams, technology selections, threat model, and capacity plan committed to the feature's .ai directory.

What is a C4 model diagram in software architecture?

The C4 model visualizes architecture at multiple levels; this skill generates Level 1 Context diagrams showing actors and external systems, and Level 2 Container diagrams showing internal services, databases, queues, and trust boundaries in Mermaid syntax.

When should I write an Architecture Decision Record?

Write an ADR for any choice where a reasonable engineer could pick differently and changing it later costs significant rework — database engines, sync vs async communication, auth mechanisms, partitioning strategies, and third-party dependencies are explicit triggers.

Can I skip the HLD for small changes or bug fixes?

Yes. Bug fixes, config changes, isolated utilities, and additions following an established service pattern are explicitly out of scope, but the skill states the skip inference and asks for confirmation rather than skipping silently.

Does the HLD process include security threat modeling?

Yes. It applies the Shostack 4-Question Frame with STRIDE per component, documents security controls for authentication, authorization, and encryption, and requires verification that all identified threats have documented mitigations.

Why can't implementation planning start right after the HLD is written?

The skill enforces a hard gate: the hld-reviewer agent must pass with no Critical or Important findings, and a human must explicitly approve the HLD before the writing-plans skill activates. Silence is not treated as approval.