design-review-prework

Generates architecture diagrams, system reports, and intake answers for Temporal design reviews from read-only codebase exploration.

Updated Sep 2, 2026
One-click install
npx skills add https://github.com/jmbarzee/temporal-design-review-prework --skill design-review-prework-jmbarzee
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: design-review-prework
Source: https://github.com/jmbarzee/temporal-design-review-prework/tree/main/skills/design-review-prework
Command: npx skills add https://github.com/jmbarzee/temporal-design-review-prework --skill design-review-prework-jmbarzee

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) and references (resource) components.

What problem does it solve? Preparing for a Temporal design review usually means scrambling to explain your system to a Solutions Architect, resulting in thin submissions that turn the meeting into generic Temporal education. This Skill builds a complete prework bundle — external and internal architecture diagrams, a system report, SA intake answers, and a specific agenda — from read-only exploration of your own codebase plus a short intake conversation. ## Core Features & Use Cases - Read-only codebase survey: Scans Temporal SDK usage, worker registrations, workflow triggers, and durability mechanics while excluding generated code, with a strict no-network, no-modification policy. - Validated Mermaid diagrams: Produces external topology and internal workflow-shape diagrams that pass a bundled linter and optional renderer, so nothing you share fails to open. - Neutral, evidence-tagged reporting: Every fact is marked observed (with file:line), stated, or effective, with a gap ledger for unknowns — the bundle maps the system without grading it. - Use Case: Before a scheduled design review, run the skill against your services to hand your Temporal SA a rendered architecture diagram set, a mechanism-and-values report, and a pre-built agenda instead of "4 boxes and lines." ## Quick Start Ask your AI coding agent to help you prepare prework for your Temporal design review, then answer the short intake questions and confirm the discovered scope.

Frequently Asked Questions about design-review-prework

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

FAQPage Schema
How do I prepare for a Temporal design review?

Run this skill against your codebase to produce the prework bundle: external and internal architecture diagrams, a system report with exact timeout and retry values, answers to the questions a Temporal Solutions Architect will ask, and a specific agenda. You review the bundle and share it yourself via the share manifest.

What does the design review prework bundle contain?

The bundle contains report.md, intake.md with stated or observed answers, a gap ledger of unknowns, a share manifest, and Mermaid diagrams for external architecture and each in-focus workflow. Rendered PNG or SVG images are included when a Mermaid renderer is available.

Does this skill modify or send my source code anywhere?

No. The skill is strictly read-only against your code and makes no network calls after an optional Phase 0 renderer install. The bundle contains diagrams and prose only, no source code, and nothing leaves your machine — you share files yourself.

Which Temporal SDKs and languages are supported?

The survey script handles Go, Python, TypeScript, and Java codebases, detecting Temporal SDK imports, worker construction, RegisterWorkflow and RegisterActivity call sites, and workflow triggers. Generated code such as .pb.go files and vendor directories is excluded from all counts.

Why do my Mermaid diagrams fail to render?

Common causes are unquoted node labels containing commas or parentheses, literal newlines in labels, HTML entities, and missing in-canvas legends. Run the bundled lint_diagrams.py script on your .mmd files; it reports each issue with a fix, and Phase 5 requires a clean pass.

What are the limitations of the automated codebase survey?

The survey cannot see deployment-time wiring, config-driven routing, or services without available source; these go into the gap ledger as open questions. It also states mechanisms and values without assessing design quality — evaluation is left to the Solutions Architect.