Design evaluation tests probing failure modes in LLM-based systems.
npx skills add https://github.com/CAPHTECH/claude-marketplace --skill lde
Please help me install this Agent Skill. Skill: lde Source: https://github.com/CAPHTECH/claude-marketplace/tree/main/caphtech-plugin/skills/lde Command: npx skills add https://github.com/CAPHTECH/claude-marketplace --skill lde
This Skill implements Law-Driven Engineering methodology, transforming business constraints into codified laws with verification and observability requirements.
Set up a new LDE project by selecting the appropriate track (Simple/Standard/Complex) and following the A-E development phases.
Initialize an LDE project for inventory management, documenting key business laws and their verification methods.
High-intent search queries and answers about installing and using this skill.
Law-Driven Engineering identifies evaluation gaps by designing tests that probe core failure modes like hallucination, prompt overfitting, and partial processing. Golden-dataset design and multi-scorer evaluation with configurable thresholds validate coverage across diverse inputs.
Law-Driven Engineering transforms business constraints into codified laws with verification and observability requirements. It ensures business-critical systems meet compliance and quality standards through systematic validation of core failure modes in LLM workflows.
Initialize an LDE project by selecting the appropriate track—Simple, Standard, or Complex—based on system scope. Follow the A-E development phases to document key business laws and their verification methods systematically.
Yes. Law-Driven Engineering applies directly to LLM-driven workflows and agents, providing assay-kit based test-case generation, failure-mode mapping, and generalization-pattern validation to ensure reliable system behavior.
Law-Driven Engineering targets hallucination, prompt overfitting, and partial processing failures. It uses multi-scorer evaluation and configurable thresholds to validate that LLM systems handle edge cases and diverse inputs correctly.
imMamdouhaboammarTransforms business inputs into concise executive summaries using McKinsey SCQA framework.
CAPHTECHDesigns LLM evaluation plans with failure-mode coverage and test-case templates.
rocklambrosEvaluate RAG pipelines with retrieval metrics, generation scores, and failure attribution.
wojonsInstrument code to trace execution and verify each component and state transition.
