A growing collection of high-leverage AI agent skills designed to embed expert intuition into your automated developer workflows.
Whether you are focusing on cybersecurity, devtools, productivity, code review, or system architecture, these skills force AI agents to think procedurally and rigorously by relying on structured outputs and task-specific evaluation criteria rather than generic guidance.
| Skill | Category | Description | Ecosystem |
|---|---|---|---|
| vibe-security-pentest | Cybersecurity |
Embed a penetration tester's mindset into every phase of AI-assisted development (vibe coding). Prevent race conditions, IDOR, logic flaws, and insecure file uploads. Inspired by the Yuri Dev experiment (1 hacker vs 4 vibe coders). | skill.sh |
(More skills spanning frontend architecture, backend performance, and deployment automation coming soon...)
Add a skill to your project workspace globally or locally:
npx skills add https://github.com/nullfrank/skills --skill <skill-slug>For example, to install the vibe-security-pentest skill:
npx skills add https://github.com/nullfrank/skills --skill vibe-security-pentestThese skills are designed specifically for the rapidly evolving world of vibe coding and AI-assisted software engineering.
- Progressive Disclosure: Relevant instructions and data are loaded only when conceptually triggered, saving AI context window limits.
- Specific Output Formats: No empty buzzwords; every prompt expects concrete verification or generated assertions.
- Task-Specific Mastery: We prefer deep, narrow skills over bloated, multi-domain "experts".
Pull requests and issues will be reviewed as the repository scales to support more capabilities. If you have an idea for a skill that solves a painful, repeated task with a concrete output, feel free to open a discussion!
Maintained by @nullfrank.