Constrain LLM outputs to valid JSON, XML, and code using regex and grammars.
npx skills add https://github.com/kotakbiasa/hermes-agent --skill guidance-kotakbiasa
Please help me install this Agent Skill. Skill: guidance Source: https://github.com/kotakbiasa/hermes-agent/tree/main/optional-skills/mlops/guidance Command: npx skills add https://github.com/kotakbiasa/hermes-agent --skill guidance-kotakbiasa
Guidance provides a constrained generation framework to enforce structure and correctness in LLM outputs using regexes, grammars, and workflow-driven prompts.
Install Guidance and wrap prompts with system, user, and assistant contexts to generate constrained outputs.
High-intent search queries and answers about installing and using this skill.
Regex and grammar-based generation enforces structured outputs by constraining the LLM to predefined formats. This ensures validity for JSON, XML, code, and structured data workflows across local and API backends.
Token healing is a constrained generation technique that fixes token boundary issues during multi-step workflows, ensuring LLM outputs strictly adhere to predefined regex patterns and grammars without structural errors.
Yes, grammar-based constrained generation supports both local and API backends, allowing you to apply regex constraints and Pythonic tooling to produce auditable, repeatable structured data outputs.
You build multi-step workflows by wrapping prompts with system, user, and assistant contexts, then applying guidance-style orchestration and token healing to execute robust, structured generation tasks.
Use regex constraints for LLM generation when you need guaranteed format validity for structured data, code, or multi-step workflows where unguided outputs would break downstream processing or validation.
Yes, constrained generation integrates with Pythonic tooling to wrap prompts and apply grammars, creating repeatable, auditable workflows that reliably produce valid structured outputs.
ovachieverConstrain LLM outputs to valid JSON, XML, and code using regex and grammar constraints.
dojoengineConfigure Dojo world ownership, namespaces, and resource permissions via deployment TOMLs and WorldStorage API.
beratcelik1Orchestrate autonomous Claude Code loops for multi-agent pipelines and continuous iterations.
qcmuuServe LLM inference with RadixAttention prefix caching and structured output constraints.
