How does this skill handle AI quality degradation?

The skill enforces 'Atomic Scoping,' splitting projects into many small plans. It monitors context usage and triggers a handoff or pause at 40-50% usage, before Claude enters 'completion mode' where quality typically drops.

What makes a plan 'Claude-executable'?

A Claude-executable plan (PLAN.md) acts as a direct prompt. It includes specific file references, measurable success criteria, and atomic tasks that Claude can perform via CLI tools without needing manual guidance.

Does this skill work with existing repositories?

Yes, it performs a context scan upon invocation to detect existing git repos, planning structures, and BRIEF.md files, allowing you to resume or extend existing projects seamlessly.

What are human checkpoints in this workflow?

Checkpoints are reserved for visual verification (like UI checks) or strategic decisions. If a task can be done via CLI (like deployments, tests, or builds), the skill mandates that Claude automates it.

Agentic Project Architect

Name: Agentic Project Architect
Author: shishiv

byshishiv

生产力与工作流

Generates hierarchical, execution-ready project plans optimized for solo development with Claude.

关于

This skill streamlines the planning process by creating atomic, executable project plans tailored specifically for a single developer working with Claude. It focuses on maintaining high output quality by aggressively managing context usage, breaking complex phases into small, verifiable tasks, and prioritizing CLI-driven automation. By replacing enterprise-heavy documentation with 'plans-as-prompts,' it ensures that every roadmap, phase plan, and task list is designed for immediate action and clear verification without coordination overhead.

主要功能

CLI-first automation protocol to minimize manual human intervention
Domain-specific expertise loading for macOS, iOS, Unity, and web development
Context-aware monitoring with automated handoffs at 50% usage
Built-in deviation rules for automatic bug fixing and blocker resolution
0 GitHub stars
Atomic task planning (2-3 tasks per plan) to prevent AI quality degradation

使用场景

Managing long-running development tasks with context-safe handoffs
Scaffolding a new project from a high-level brief into a technical roadmap
Breaking down complex development phases into small, surgical implementation plans

关于

主要功能

CLI-first automation protocol to minimize manual human intervention
Domain-specific expertise loading for macOS, iOS, Unity, and web development
Context-aware monitoring with automated handoffs at 50% usage
Built-in deviation rules for automatic bug fixing and blocker resolution
0 GitHub stars
Atomic task planning (2-3 tasks per plan) to prevent AI quality degradation

使用场景

Managing long-running development tasks with context-safe handoffs
Scaffolding a new project from a high-level brief into a technical roadmap
Breaking down complex development phases into small, surgical implementation plans