AI Engineering

Generates high-quality cover images, inline illustrations, and infographics for WeChat public account articles. Supports AI-generated visuals and HTML-rendered precise graphics, with an interactive design proposal workflow.

Claude Code Skill Builder orchestrates four specialized agents to elicit requirements, generate files, validate structure and code, and enhance documentation, delivering production-ready Claude Code skills.

WebMCP implements WebMCP in web projects by adding browser-native structured tools for AI agents using imperative JavaScript or declarative HTML APIs. Use it for setup, tool creation, debugging, auditing, and testing.

Belief Duel resolves contradictions in an AI's memory graph when two accepted beliefs cannot logically coexist. Use it when sibling memories conflict or a child node challenges a parent node's default belief.

Memory Audit is an entry point for auditing an AI agent's memory. It routes symptoms like stale or bloated memories to sub-skills for pruning, extracting patterns, and resolving contradictions.

Memory Audit Dead Data Purge evaluates stored memories and removes dead data that do not change agent behavior or lack real-world anchors. Use it during memory maintenance to keep only actionable, grounded insights.

Memory Audit Node Decomposition breaks an overstuffed memory node into focused independent nodes when a single disclosure cannot cover all its concepts. Use it when a memory exceeds 800 tokens, mixes unrelated concepts, or triggers jump-reading.

Memory Audit Pattern Extraction reviews an agent's memory network for repeated failures, identifies why prior 'antidote' memories did not prevent recurrence, and rewrites them with stronger triggers and positive mindset anchors. Use it when several memories describe the same lesson or the same mistake recurs.

Memory Discoverability Audit reviews an AI agent's stored memories to verify parent placement, disclosure triggers, aliases, and priority so each memory surfaces at the right moment. Use it when a memory fails to surface, trigger conditions are vague, or a parent has more than ten children.

Agent Workspace Init bootstraps a new multi-agent repository from the RepoBrain template. It supports quick scaffolds and full runtime profile setup with MCP toggle, swarm preference, sandbox type, and optional git init.

Graph Retrieval queries a normalized graph store to return semantic triples and evidence for codebase structure and dependency questions. Use it when an agent needs replayable, context-enriching graph data without bypassing the existing pipeline.

Knowledge Layer provides a high-level deployment wrapper over RepoBrain core with graph-first knowledge injection and all-file support. Use it to build and query a repository knowledge graph via refresh_filesystem and ask_filesystem.