Agent Skills by lightxue
Showing 23 vetted skills indexed across 1 GitHub repositories.
ljg-writes
Refine a writer's central viewpoint into a coherent argument from draft.
using-git-worktrees
Create isolated git worktrees with safety checks and automatic project setup.
ljg-word
Explains a single English word's semantics, etymology, and usage in structured Markdown.
test-driven-development
Write failing tests before implementation using the red-green-refactor cycle.
dom-console
Generate JavaScript commands from DOM HTML fragments for Chrome DevTools Console execution.
systematic-debugging
Identify and fix root causes using a four-phase debugging workflow.
using-superpowers
Enforce pre-response skill discovery and invocation with priority rules.
dispatching-parallel-agents
Dispatch parallel agents to independent tasks without shared state.
deep-research
Retrieve and validate multi-source evidence to build citation-backed research reports.
ljg-rank
Identify minimal independent generators underlying a domain and validate four criteria.
executing-plans
Coordinate written plan execution with review checkpoints and verifications.
finishing-a-development-branch
Validate tests and offer merge, PR, keep, or discard options for a development branch.
ljg-learn
Deconstruct concepts into eight exploratory dimensions and produce a markdown report.
ljg-paper
Summarize research papers from arXiv links or PDFs into plain-language takeaways.
brainstorming
Guide collaborative brainstorming to produce an approved design document before implementation.
writing-plans
Convert product requirements into implementation plans with tasks and milestones.
requesting-code-review
Route code changes through a dedicated code-review subagent using a structured template.
receiving-code-review
Restate code review feedback, verify against the codebase, and implement changes one at a time.
writing-skills
Create, edit, and validate Skills documentation with YAML frontmatter.
spec-workflow
Automate four-phase development workflows with validation scripts and auditable wiki artifacts.
verification-before-completion
Require fresh command output and exit codes before declaring success.
subagent-driven-development
Coordinate implementation plans by deploying subagents with isolated context and staged reviews.
excel-formula
Generate Excel formulas from natural language descriptions with explanations.