Constrain LLM outputs with regex and grammars for valid structured data.
npx skills add https://github.com/Harries/hermes-agent --skill guidance-harries
Please help me install this Agent Skill. Skill: guidance Source: https://github.com/Harries/hermes-agent/tree/main/optional-skills/mlops/guidance Command: npx skills add https://github.com/Harries/hermes-agent --skill guidance-harries
Guidance enables precise control over LLM outputs by applying regex constraints and grammars, ensuring valid structured data and repeatable results across tasks.
Create a constrained JSON object with a name and age validated by regex.
High-intent search queries and answers about installing and using this skill.
You can constrain LLM outputs to guarantee valid JSON by applying regex constraints and grammars. This ensures deterministic structured data generation, preventing formatting errors and invalid syntax in the output.
Token healing is a feature that fixes tokenization boundary issues during constrained generation. It automatically adjusts token boundaries to ensure regex and grammar constraints apply smoothly without breaking the intended text structure.
Yes, constrained generation works across both local and API LLM backends. The underlying library loads constraints and grammars from references, enabling deterministic structured data generation regardless of the backend environment.
The best way to extract structured data is by defining grammars or regex patterns to constrain the LLM output. This approach guarantees valid structured formats like JSON or XML directly from unstructured text inputs.
Yes, constrained generation supports multi-step workflows and pipelines through Pythonic composition. You can chain multiple constrained generation steps together to build complex agents or data processing pipelines.
AissenLiuConstrain LLM outputs with regex and grammar constraints.
ovachieverConstrain LLM outputs to valid JSON, XML, and code using regex and grammar constraints.
eddieluengEnforce regex and grammar constraints for valid JSON, XML, and code outputs.
thejordanleopoldDiagnose server-side bottlenecks in API latency, database queries, and memory usage.
