Agentic Development Vision
Overview
This document outlines our vision and strategy for AI-assisted development in the Cloud Control Plane repository.
Philosophy
TODO: Define our approach to human-AI collaboration
Available Agents
The orchestrated Shopify/general-dev agents (css-implementer, css-verifier,
code-implementer, code-verifier, search-tuner, search-verifier,
qc-investigator, finisher) are documented in their respective skill READMEs
(.claude/skills/integrate-storefront, .claude/skills/general-feature-pr,
etc.). See docs/dev/orchestration/followups.md for orchestration design notes.
Deferred — reserved for a future plan-first-feature-pr workflow
| Agent | Purpose | Status |
|---|---|---|
plan-verifier | Validate plans and proposals before implementation | Deferred; see docs/dev/orchestration/followups.md #5 |
test-case-generator | Generate comprehensive test-case markdown documentation (not test code) | Deferred; see docs/dev/orchestration/followups.md #5 |
For clarification of vague specs, use the /grill-me skill (canonical
clarification entry point) rather than a separate interviewer agent.
Skills
TODO: Document available skills
Best Practices
TODO: Guidelines for effective prompting and agent usage
Roadmap
TODO: Future capabilities and improvements