Guidance
Skill Verified ActiveControl LLM output with regex and grammars, guarantee valid JSON/XML/code generation, enforce structured formats, and build multi-step workflows with Guidance - Microsoft Research's constrained generation framework
To provide developers with fine-grained control over LLM output, ensuring validity, structure, and reducing generation latency for tasks requiring precise formatting and multi-step reasoning.
Features
- Constrained LLM output generation via regex and grammars
- Guaranteed valid JSON, XML, and code generation
- Token healing for natural continuations
- Support for multiple LLM backends (API and local)
- Pythonic control flow for multi-step workflows
- Selection constraints for fixed choices
Use Cases
- Generating valid JSON or XML outputs
- Enforcing specific data formats (emails, dates, phone numbers)
- Building structured agent workflows with decision trees
- Reducing LLM latency by constraining generation paths
- Creating reproducible LLM outputs through precise grammars
Non-Goals
- Replacing the core LLM inference engine
- Providing a low-level interface for raw token generation
- Handling the execution of external tools (delegated to user code)
- Managing LLM agent orchestration beyond generation control
Execution
- info:Pinned dependenciesWhile dependencies are listed, specific version pinning and lockfiles are not explicitly detailed in the SKILL.md, relying on standard pip behavior.
Installation
First, add the marketplace
/plugin marketplace add Orchestra-Research/AI-Research-SKILLs/plugin install AI-Research-SKILLs@ai-research-skillsQuality Score
VerifiedTrust Signals
Similar Extensions
Guidance
79Control LLM output with regex and grammars, guarantee valid JSON/XML/code generation, enforce structured formats, and build multi-step workflows with Guidance - Microsoft Research's constrained generation framework
Teach Guidance
100Guide a person in becoming a better teacher and explainer. AI coaches content structuring, audience calibration, explanation clarity, Socratic questioning technique, feedback interpretation, and reflective practice for technical presentations, documentation, and mentoring. Use when a person needs to present technical content and wants preparation coaching, wants to write better documentation or tutorials, struggles to explain concepts across expertise levels, is mentoring a colleague, or is preparing for a talk or knowledge-sharing session.
Moyu (摸鱼)
100과잉 엔지니어링 패턴이 감지되면 자동으로 활성화됩니다: (1) 사용자가 명시적으로 변경을 요청하지 않은 코드나 파일을 수정하는 경우 (2) 요청되지 않은 새로운 추상화 레이어(class, interface, factory, wrapper)를 생성하는 경우 (3) 요청되지 않은 주석, 문서, JSDoc, 타입 어노테이션을 추가하는 경우 (4) 요청되지 않은 새로운 의존성을 도입하는 경우 (5) 최소한의 편집 대신 파일 전체를 다시 작성하는 경우 (6) diff 범위가 사용자의 요청을 명백히 초과하는 경우 (7) 사용자가 "너무 많아", "거기는 건드리지 마", "X만 변경해", "간단하게", "그만" 등의 신호를 보내는 경우 (8) 발생할 수 없는 시나리오에 대한 에러 처리, 유효성 검사, 방어적 코드를 추가하는 경우 (9) 요청되지 않은 테스트, 설정 스캐폴딩, 문서를 생성하는 경우
Create Command
100Interactive assistant for creating new Claude commands with proper structure, patterns, and MCP tool integration
Write A Skill
100Create new agent skills with proper structure, progressive disclosure, and bundled resources. Use when user wants to create, write, or build a new skill.
Openclaw
100Generate images and videos from text with multi-provider routing — supports GPT Image 2.0 (near-perfect text rendering), Nanobanana 2, Seedream 5.0, Midjourney V8.1 (unified photorealistic + anime), Flux 2 Klein (cheap drafts), Seedance 2.0 / Happyhorse 1.0 / Veo 3.1 video, and local ComfyUI workflows. Includes 1,446 curated prompts and style-aware prompt enhancement. Use when users want to create images/videos, design assets, animate photos, enhance prompts, or manage AI art workflows. NOT for: generic chat, code generation, document writing, video editing of existing footage, audio/TTS, or any task unrelated to AI image/video creation.