665,000+ skills indexed

HankLiu

Community

@hankliu447

1Followers

|

2Public Repos

|

21Published Skills

劉璟融(Hank) · Founder @ FlyingHighMedia ⁠Building FlySuite Ontology-driven SaaS for commerce Rust · Go · PostgreSQL 成大 AMBA 📍 Tainan, Taiwan

🐙 GitHub ↗

Agent Skills by HankLiu

Showing 21 vetted skills indexed across 1 GitHub repositories.

superspec:kickoff

Guide a four-phase flow to produce plan.md, specs, and tasks.

spec-validation

Validate specification structure, content, cross-references, and completeness.

systematic-debugging

Diagnose root causes of technical issues through a four-phase debugging workflow.

gemini

Run Gemini CLI for code analysis and generation with custom prompts.

phase-protocol

Enforce entry and exit gates with durable TODO anchoring across development phases.

superspec:archive

Apply delta changes to main specs and archive documents after verification.

superspec:brainstorm

Convert ideas into proposal.md, design.md, and specs/*.md artifacts.

superspec:verify

Load specs, map scenarios to tests, and report coverage gaps.

dispatching-parallel-agents

Dispatch dedicated agents to investigate independent problem domains in parallel.

code-review

Automate two-stage code reviews for specification compliance and code quality.

codex

Automate code analysis, refactoring, and edits via Codex CLI.

executing-plans

Execute plan tasks in batches with TDD and review checkpoints.

superspec:plan

Generate plan.md and tasks.md from validated specs with test-first tasks.

external-review

Route frontend and backend code reviews to Gemini or Codex via superspec/project.yaml.

superspec:finish-branch

Present four branch completion options after tests pass.

git-worktree

Initialize isolated git worktrees for feature work with deterministic directory selection.

using-superspec

Locate and invoke SuperSpec skills via Claude Code at conversation start.

receiving-code-review

Guide engineers to verify code-review feedback against the codebase before implementing changes.

superspec:execute

Orchestrates plan execution by dispatching subagents and applying two-stage reviews.

verification-before-completion

Require verification commands and evidence before any completion claim.

tdd

Guide red-green-refactor cycles by writing failing tests before code.