Constrain LLM outputs with regex and grammars for valid JSON, XML, or code.
npx skills add https://github.com/AlvaroBiano/hermes-agent --skill guidance-alvarobiano
Please help me install this Agent Skill. Skill: guidance Source: https://github.com/AlvaroBiano/hermes-agent/tree/main/optional-skills/mlops/guidance Command: npx skills add https://github.com/AlvaroBiano/hermes-agent --skill guidance-alvarobiano
Guidance provides a framework to control LLM outputs using regex and grammars, ensuring JSON/XML/code generation is valid, and enabling multi-step workflows with Pythonic control flow.
Ask the AI to generate a constrained JSON object using a regex and grammar to guarantee valid syntax.
High-intent search queries and answers about installing and using this skill.
You can constrain LLM outputs to generate valid JSON by applying regex and grammar-based constrained generation. This guarantees syntactically valid structured outputs without needing post-processing validation.
Token healing is a mechanism that fixes tokenization artifacts at pattern boundaries during constrained generation. It ensures smooth transitions between static text and generated content in multi-step workflows.
Yes, constrained generation works with local models using transformers and llama.cpp backends. This enables private, fast inference while maintaining valid JSON, XML, or code generation.
Yes, you can build multi-step workflows using Pythonic control flow and guidance decorators. This allows complex generation pipelines where each step's output feeds into the next constrained generation task.
The best way to guarantee valid syntax for data extraction and form validation is using grammar-based constrained generation. It enforces output formats like JSON or XML directly during inference, eliminating invalid outputs.
Yes, you need the guidance and transformers dependencies installed to implement constrained generation pipelines. These backends provide the infrastructure for grammar-based output control and local model inference.
ricableGenerate grammar-constrained JSON, XML, regex, and code outputs.
ovachieverConstrain LLM outputs to valid JSON, XML, and code using regex and grammar constraints.
Clay-HHKEnforce token-level regex and grammar constraints for structured LLM outputs.
heysuhasConstrain LLM token sampling to JSON schemas, regex patterns, or Pydantic models.
