AI Engineering

Gemini runs Gemini CLI for code review, plan review, and large-context analysis. Use it when you need to inspect many files or work with contexts over 200k tokens.

Image Prompting writes ready-to-use prompts for Nano Banana and GPT Image 2. Use it when you need an image prompt, image edit, vision analysis, or model-specific settings like quality and size.

Knowledge Graph maintains an append-only entity store with atomic facts and short summaries. Use it to remember durable context about people, companies, and projects across sessions.

Munger Observer reviews a specific decision, plan, or artifact using bounded evidence, counterevidence, alternatives, and verifiable next checks. Use it for decision reviews, premortems, and blind-spot reviews without assessing a person’s character or mental state.

My Router detects the task type from prompts, file types, and project hints, then dispatches to the right skill. It is used when a repo needs context-aware skill selection or a local routing table.

Remotion makes videos from notes, articles, scripts, or product briefs and renders them to MP4. Use it when creating product demos, promos, trailers, explainers, motion graphics, or animated logos.

Self-Improving Agent captures durable learnings, errors, and feature requests in a local knowledge base. Use it before major tasks or after failures and corrections so the same issue is less likely to repeat.

Senior Engineering Principles guides planning, implementation, and review for non-trivial software work. Use it when architecting systems, breaking down tasks, adding quality gates, or coordinating multi-agent builds.

Video helps create, improve, audit, and split prompts for AI video generators, including storyboards, shot lists, camera direction, continuity, and sound design. Use it when turning a scene or script into prompts for tools like Kling, Veo, Seedance, Luma, Pika, or Sora.

MUE activates a self-evolving AI agent mode with setup, status, evolution, and absorption commands. Use it to configure and run the agent in a chosen domain.

Value For Fable applies Fable 5 operating rules to a Sonnet session for outcome-first communication, verification discipline, parallel tool use, and concise final answers. Use it for code, writing, research, or documentation work that should follow those session constraints.

Adversarial Review Orchestration fans out findings, then verifies each one through multiple skeptical perspectives before accepting conclusions. Use it for security audits, code reviews, research synthesis, or migrations when one agent’s self-report cannot be trusted.