AI Engineering

Virtual Companion adds a virtual companion to OpenClaw that chats proactively during tasks, generates scene photos, and sends warm messages via Feishu.

Xiaoyue Companion generates warm conversational responses via the Zhipu AI API and optionally sends static images. Use it when users express tiredness, need encouragement, or want a friendly check-in during tasks.

Tool Calling Tutor walks users through diagnosing and fixing tool-calling problems when an LLM won't call a tool, passes wrong arguments, or ReAct loops stall. It uses a 4-branch diagnostic and a 5-step schema design walkthrough, with references to bad/good schema examples and SDK differences.

Agentic Workflows routes requests to design, create, update, debug, or upgrade agentic workflows to the correct prompt files. Use it when a user asks to build or modify agentic workflows in a repository, and follow the matching prompt or skill.

Create Custom Agent scaffolds VS Code custom agent files (.agent.md) that define specialized AI personas with tools, instructions, and handoffs. Use it when creating new agents, configuring workflows, or setting up agent-to-agent handoffs.

Skill Scaffolder creates new agent skills that conform to the Agent Skills specification and dotnet/skills conventions, generating SKILL.md frontmatter, section templates, and validation guidance. Use it when setting up a new skill directory or writing a description that the runtime will route to.

Technology Selection helps developers choose the right .NET AI/ML stack (ML.NET, Microsoft.Extensions.AI, Microsoft Agent Framework, ONNX Runtime, OllamaSharp) for classification, LLM integration, RAG, agentic workflows, and local inference in .NET 8+ projects.

Agent Identifier helps you create and configure Claude Code agents, including YAML frontmatter, triggering descriptions, system prompts, model, color, and tool settings. Use it when setting up autonomous subagents, writing agent frontmatter, or designing agent behavior.

Architecture Design defines the standard code architecture for machine learning projects using Factory and Registry patterns. Use it when creating new registrable ML components such as datasets, models, or modules that require decorators and factory wiring.

Command Development guides creating Claude Code slash commands, covering Markdown structure, YAML frontmatter, dynamic arguments, file references, and bash execution. Use it when building reusable, interactive, or namespaced commands for agent workflows.

Hook Development provides guidance for creating Claude Code plugin hooks, focusing on prompt-based hooks and event-driven automation. Use it when implementing PreToolUse, PostToolUse, Stop, SubagentStop, SessionStart, or other hook events to validate tool use, enforce policies, or add context.

MCP Integration guides configuring Model Context Protocol servers in Claude Code plugins to expose external service capabilities as tools. Use it when adding, integrating, or connecting MCP servers (SSE, stdio, HTTP, WebSocket) with authentication and environment variable configuration.