Traction scoreGitHub stars can be faked, so popularity alone can be misleading. Traction Score looks for broader signs of recent attention, adoption, and active maintenance.
This collection of Claude Code skills covers the full development cycle: brainstorming designs, executing plans in batches with review checkpoints, systematic debugging root cause first, test-driven development, code review, and problem-solving techniques like inversion and simplification cascades. It also includes meta-skills for writing, testing, and contributing new skills upstream. The library serves developers who want their AI assistant to follow disciplined workflows rather than improvise.
WHO IT'S FOR
AI-Assisted Software Engineers
use skills to accelerate AI-assisted development
Technical Leads and Architects
apply structured techniques for design and trade-offs
QA and Test Automation Engineers
apply rigorous debugging and testing methodologies
Compatible AgentsThe repository documents support for these agents. The skills may also work with other agents that can load SKILL.md files, but they may need some setup or small changes.
Follow the documented setup, then try a first task.
Claude Code superpowers plugin
Installs: Complete Superpowers Skills core skills library from this repository · Claude Code
Before you start
Claude Code with the superpowers plugin already installed
Use the superpowers plugin in Claude Code. The plugin automatically clones this repository into the documented skills directory; no terminal command is provided in the supplied evidence.
Use the Inversion Exercise skill on this problem: [describe the problem]. Start from this assumption: [state one assumption that currently seems mandatory]. List the core assumptions, invert each one, briefly explore the implications, and identify any inversion that could work in this context. Keep this to a small first-pass analysis; do not implement changes.
Use the Scale Game skill to evaluate this approach: [describe the approach]. Focus on this scale dimension: [volume, speed, users, duration, or failure rate]. Test both a much smaller and a much larger extreme, then report what breaks, what survives, and one architecture question the exercise reveals. Do not modify the implementation.
The observable condition that signals completion, if known
Use the Condition-Based Waiting skill to review this flaky asynchronous test: [paste the test or provide its file path]. Identify one arbitrary delay that is not testing timing behavior, propose a condition-based replacement using the test's language and framework, and explain the timeout and polling condition. Return only a focused suggested patch; do not apply it.
Claude