umputun/cc-thingz

What it solves

It provides a curated marketplace of plugins for Claude Code, adding specialized capabilities that the base tool lacks, such as structured implementation planning, rigorous code review workflows, and collaborative design dialogues.

How it works

The project organizes a set of independent plugins (hooks, skills, and commands) that can be installed via the Claude Code plugin system. These plugins extend the AI's behavior by:

• Skills: Adding new slash commands (e.g., /brainstorm, /review) that trigger multi-phase AI workflows.
• Commands: Creating structured files (like implementation plans) and managing them.
• Hooks: Intercepting specific events (like ExitPlanMode or UserPromptSubmit) to trigger external scripts or editor overlays for human-in-the-loop validation.
• Subagents: Utilizing Claude Code's subagent capabilities to perform parallel reviews (quality, testing, documentation) and autonomous task execution.

Who it’s for

Developers using Claude Code who want to move beyond simple chat interactions toward a professional software engineering lifecycle, including formal planning, autonomous execution, and strict technical writing standards.

Highlights

• Structured Planning: A /planning:make command that guides the user through context gathering and approach trade-offs before generating a formal plan file.
• Autonomous Execution: An /planning:exec skill that executes a plan task-by-task using isolated subagents and a multi-phase review process.
• Interactive Review Overlays: Integration with terminal overlays (agterm, tmux, kitty, wezterm) to allow users to annotate plans or diffs directly in their editor, which the AI then uses to revise its work.
• Deep PR Analysis: A review tool that detects scope creep, analyzes architecture, and uses git worktrees to avoid disrupting the local environment.
• Analytical Toolset: Specialized skills for dialectic analysis, root cause investigation, and a "writing-style" guide to eliminate AI-speak from technical communication.