domains / families / software-engineering / llm-app-patterns

llm-app-patterns

authentic

Architecture and integration sketches for LLM applications, with explicit retrieval, tool, privacy and verification boundaries.

When to use

• Designing LLM-powered applications
• Implementing RAG (Retrieval-Augmented Generation)
• Building AI agents with tools
• Setting up LLMOps monitoring
• Choosing between agent architectures

When not to use

• A retrieved source list does not prove each answer claim is supported; verify claim-to-source evidence and abstention behavior.
• Prompt text and JSON-shaped output are not authorization boundaries. Enforce permissions in the application.
• Provider failover can change output quality, tool schemas, cost and data residency; only use pre-approved compatible fallbacks.
• Caching must respect tenant access, data/prompt revisions and deletion policy; temperature zero does not make output deterministic.
• Model/SDK examples are not a complete service, benchmark or production-readiness certificate.

Skill page

Source