Agent Skills by liauw-media
Showing 35 vetted skills indexed across 2 GitHub repositories.
system-architect
Audit system configurations and enforce defense-in-depth security controls.
ci-templates
Provide standardized CI/CD templates for GitHub Actions and GitLab CI.
branch-discipline
Create one Git branch per issue with small, focused commits.
git-worktrees
Create isolated git worktrees for parallel feature development in one repository.
git-workflow
Enforce atomic Git commits with pre-commit verification and standardized messages.
frontend-design
Develop distinctive frontend interfaces with production-grade code and cohesive aesthetics.
brand-guidelines
Establish brand guidelines stored in .claude/BRAND-GUIDELINES.md and .claude/brand/ assets.
pre-commit-hooks
Configure pre-commit hooks for JavaScript/TypeScript, Python, and PHP projects.
resource-limiting
Limit CPU and I/O usage with nice, ionice, and cpulimit on shared servers.
git-platform-cli
Create issues and pull requests via gh or glab across GitHub and GitLab.
remote-code-agents
Delegate long-running code tasks to remote Claude Code agents via API or CLI.
subagent-driven-development
Dispatch fresh subagents sequentially for plan-driven tasks with mandatory reviews.
dispatching-parallel-agents
Dispatch parallel agents to execute independent tasks concurrently.
finishing-a-development-branch
Automate and enforce a pre-merge checklist for Pull Requests or Merge Requests.
playwright-frontend-testing
Generate Playwright end-to-end tests from accessibility-tree data.
lighthouse-performance-optimization
Run Lighthouse audits through MCP to identify website performance bottlenecks.
browser-automation-debugging
Automate Chrome DevTools tasks for debugging and visual regression workflows.
using-skills
Detect applicable skills in the .claude/skills directory before starting tasks.
code-review
Enforces a self-review protocol covering security, performance, error handling, code structure, and tests before commit or pull request.
executing-plans
Execute multi-step plans one task at a time with verification checkpoints.
brainstorming
Conduct structured brainstorming to resolve requirement misalignment with a four-step protocol.
writing-plans
Break down brainstormed concepts into actionable implementation plans with dependencies and verification checkpoints.
requesting-code-review
Generate code-review requests with commit ranges and contextual templates.
receiving-code-review
Process code review feedback through a six-step verification workflow.