Problem
Agent-assisted development becomes inconsistent when recurring work has no shared definition of done.
agent-skills
Open-source repository
Reusable agent workflows for frontend audits, accessibility passes, bug hunts, refactors, and more.
Built by Laurent Maxhuni, a full-stack developer and AI builder from Vushtrri, Kosovo. About Laurent Maxhuni
Approach
I collected focused, installable workflows that spell out scope, verification, and handoff expectations for common engineering tasks.
Implementation
The repository packages workflows as Markdown skills with explicit scope, verification gates, and handoff expectations. It can be installed with the skills CLI.
Case study
Who it serves and how it works
- Intended user
- Developers who use coding agents and want recurring work to follow a clear scope and definition of done.
- Workflow
- Choose a task-specific skill, follow its scope and steps, run the listed verification, then use its evidence-based handoff format.
- Limitations
- A skill is guidance rather than an agent runtime. It must be checked against the current repository, available tools, and the agent's own capabilities; the repository does not publish adoption or time-saved metrics.
- Evidence
- The public repository contains the installable workflow files and their stated checks. No external adoption or performance figures are claimed.