AI Engineering

Generates original, licensed music and sound design for social media videos using safe AI tools, ensuring brand-safe audio that avoids copyright claims and mutes.

Routes AI video jobs to the best-fit generative tool and writes portable, brand-matched briefs. Sits above individual tool skills, ensuring no vendor lock-in and handling AI disclosure requirements.

Plan and guide CapCut edits for short-form social videos (TikTok, Reels, Shorts) with retention-paced cuts, auto-captions, and beat-synced sound, while ensuring licensing and safety compliance. Use when cutting a reel, setting up captions, or assessing commercial safety of CapCut assets.

Performs staged adversarial analysis of civil litigation cases under Chinese law, applying the request-right basis method to examine legal reasonableness, extract disputes, and generate review reports, Mermaid structure charts, fact lists, and strategy memos.

Generates Chinese legal research reports in Markdown format using authoritative primary sources (via legal database) and secondary web sources, following IRAC/CREAC reasoning. Adapts between lawyer and in-house counsel perspectives based on user profile.

Identifies user intent and routes to the appropriate litigation skill, offering skill navigation and workflow orchestration for Chinese civil and commercial litigation.

A litigation knowledge management hub that organizes legal research, case notes, and work notes into three zones. Supports semantic retrieval, RAG querying, and storage via multiple backends (e.g., Obsidian, Feishu), with local fallback.

A router for drafting procedural documents in Chinese civil litigation, covering 8 families such as preservation, evidence, enforcement, and jurisdiction objections. Auto-loads norms, verifies laws, and generates court-ready Word documents.

Quickly modify a single setting in your agent's profile.md without re-running the full onboarding interview. Useful for updating your role, team info, risk preferences, writing style, or backend integrations.

Generates three types of STEM exercises (worked examples, guided practice, independent exercises) with Bloom's taxonomy levels, and verifies every computational answer before output to ensure accuracy.

Provides patterns for designing reliable LLM agents with tool use, including loop termination, error handling, human checkpoints, and recognizing when an agent isn't needed.

Write short, accurate instruction files for coding agents, including commands, conventions, and hard constraints. Use it when setting up or refreshing CLAUDE.md, AGENTS.md, or similar files.