Agent Skills · Development
Curated skills for Claude, Cursor and other AI agents. Copy one prompt to install.
Applies first-principles reasoning to align scope, audit code, and plan a gradual refactoring roadmap.
A developer workflow tool for independent builders and one-person teams to generate Shrimp Review skill templates, run quality checks, apply optimization suggestions, package releases, and monitor status.
Deploys a noVNC-visible, CDP-controlled remote browser so agents and users can jointly handle logins, verification, and web debugging.
Use WorkBuddy to create, configure, preview, run, and publish XApps while managing data, assets, and local STS file sync.
Provides a development problem-solving workflow with 5Why root-cause analysis, similar-issue scans, test updates, impact analysis, and learning records.
Defines shared authentication, output, error, and state rules for invoking xiaobao-cli safely.
Detects repository context, runs a six-round Chinese wizard, and previews AGENT.MD before writing.
A standard for building Agent Skills with minimal YAML, embedded config, one execution path, tested endpoints, and root-level SKILL.md packaging.
Runs a local 2D desktop pet in a screen corner, controlling status, roles, reminders, and Claude Code/Codex session notifications via natural-language commands.
A code review guide for web and H5 frontend code, covering React/Vue, TypeScript, styles, performance, security, accessibility, and testing.
Guidance for GPU shader development, optimization, and reverse engineering across HLSL, GLSL, MSL, and WGSL, including NVIDIA cache analysis and game PAK extraction.
A key management vault with AES-256 storage, agent injection, leak scanning, log redaction, and delivery gates for multi-agent projects.