劉昱廷
Community@GongYuanCaiJi
Claude Octopus plugin providing multi-provider AI orchestration, structured development workflows, and quality gates for Claude Code users.
Agent Skills by 劉昱廷
Showing 63 vetted skills indexed across 1 GitHub repositories.
GongYuanCaiJiskill-finish-branch
Automates branch completion by running tests, reviewing diffs, and creating pull requests.
GongYuanCaiJiskill-pressure-test
Interrogates plans and designs one question at a time to expose flaws before commitment.
GongYuanCaiJiskill-staged-review
Validates implementations against spec contracts and code quality in a two-stage review pipeline.
GongYuanCaiJisys-configure
Redirects configuration requests to the interactive Claude Octopus setup wizard.
GongYuanCaiJiskill-debate
Orchestrates structured multi-round debates between Claude and external AI provider CLIs with synthesis.
GongYuanCaiJiskill-tdd
Enforces red-green-refactor test-driven development with adversarial multi-provider test review.
GongYuanCaiJiskill-writing-plans
Creates zero-context implementation plans with bite-sized TDD tasks for multi-step features.
GongYuanCaiJiflow-parallel
Decompose compound tasks into parallel work packages executed by independent claude -p processes.
GongYuanCaiJiskill-status
Displays current workflow phase, roadmap progress, blockers, and next-action suggestions from project state files.
GongYuanCaiJiskill-copilot-provider
Integrates GitHub Copilot CLI as a zero-cost provider in multi-LLM orchestration workflows.
GongYuanCaiJiskill-audit
Audits codebases systematically for broken features, inconsistencies, and quality issues with prioritized findings.
GongYuanCaiJiflow-discover
Orchestrates multi-provider AI research and synthesizes findings into a structured report.
GongYuanCaiJioctopus-architecture
Generates system architecture designs through multi-AI consensus using the backend-architect persona.
GongYuanCaiJiskill-parallel-agents
Orchestrates parallel multi-provider AI workflows for research, development, review, and delivery.
GongYuanCaiJiskill-security-framing
Validates URLs and wraps untrusted external content in security frames before analysis.
GongYuanCaiJiskill-resume
Restores prior session context from persistent workflow state files after context resets.
GongYuanCaiJiskill-doctor
Diagnoses provider, auth, config, hook, and scheduler issues across 14 environment check categories.
GongYuanCaiJiskill-code-review
Reviews code changes using multi-provider AI consensus with security analysis and PR comment posting.
GongYuanCaiJiskill-debug
Diagnose bugs through a four-phase root cause investigation and hypothesis testing workflow.
GongYuanCaiJiskill-content-pipeline
Extract structural and psychological patterns from URLs into reusable content anatomy guides.
GongYuanCaiJiskill-prd
Writes structured product requirement documents with clarification questions and adversarial multi-provider review.
GongYuanCaiJiskill-visual-feedback
Analyzes screenshot-based UI feedback to locate and fix visual issues in frontend code.
GongYuanCaiJiskill-agent-topology
Audits multi-agent workflows to determine whether each agent boundary earns its coordination cost.
GongYuanCaiJiskill-thought-partner
Facilitates structured brainstorming sessions using pattern spotting and paradox hunting techniques.
Frequently Asked Questions About 劉昱廷
FAQPage SchemaWhat tasks can I accomplish with GongYuanCaiJi's Claude Octopus skills?▼
You can orchestrate multi-provider research, debates, and code reviews; run TDD with coverage audits; perform OWASP security audits and red teaming; plan features via PRDs and NLSpec; debug systematically; manage git branch completion, rollbacks, and issue triage; and export markdown to DOCX, PPTX, XLSX, or PDF deliverables.
Who is the target user for these Claude Octopus skills?▼
Software engineers and technical leads using Claude Code who need enforced quality gates, multi-model consensus, and structured phases for features, migrations, and releases. It also suits knowledge workers producing research syntheses, design systems, slide decks, and business documents from the same environment.
How do the skills execute at runtime?▼
Skills are invoked via slash commands like /octo:debate or natural-language triggers, then run through orchestrate.sh with enforced execution modes. Pre-execution contracts and validation gates — such as synthesis files, review outputs, and coverage checks — must pass before a phase is considered complete.
What providers and cost controls are supported?▼
The manifest supports multiple external providers including a zero-cost GitHub Copilot CLI seat via copilot -p programmatic mode. Provider-health reports availability, auth, and version drift, while cost-projection and model-cost-compare skills forecast spend, warn on budget overruns, and map tasks to the cheapest adequate model.
What prerequisites or dependencies do these skills require?▼
Skills require Claude Code v2.1.12+ for forked context, task management, and agent integration, plus configured external providers set up through the /octo:setup wizard. Document export skills depend on the document-skills plugin, and git-based skills operate on repository paths with passing tests before merge.