Agent skill · product strategy · sickn33
llm-app-patterns
Architecture and integration sketches for LLM applications, with explicit retrieval, tool, privacy and verification boundaries.
Why this skill is useful
Provides executable code snippets and architectural patterns for building LLM applications, including RAG pipelines and vector database integration.
What it needs
About 2k tokens when loaded. Last updated 2026-08-07. 44,571 stars on the source repository.
What this skill does
🤖 LLM Application Patterns Architecture and integration sketches for LLM applications, with explicit retrieval, tool, privacy and verification boundaries. Detailed Guide Read the detailed guide before executing this skill. It retains the complete procedure and reference material. Treat its safety, prerequisites, and validation requirements as mandatory. For focused work, load the relevant sections; for end-to-end work, read the guide completely. When to Use This Skill Use this skill when: Designing LLM-powered applications Implementing RAG (Retrieval-Augmented Generation) Building AI agents with tools Setting up LLMOps monitoring Choosing between agent architectures --- Inputs, worked example and verification Record provider/SDK versions, authorized data and tools, request/response schemas, latency/cost budgets and the expected task outcome. All code above is an integration sketch: llm, database, parser and provider-response adapters are project-owned and must be implemented explicitly. Never execute model-provided Python, expressions or fuzzy tool names; dispatch only exact registered tools after schema and authorization checks. Enforce per-call deadlines as well as loop limits. Example: a planner first queues steps A and B, then after A replaces the remaining work with C. The executor must run A then C, never stale B. A parser step must apply the parser to the actual prior output. Two tenants with the same prompt must produce different cache keys, and cache reads/writes require an explicit reuse policy. These are concrete regression assertions; successful mocks still do not prove live provider behavior. The limiter sketch is single-threaded and process-local. A distributed deployment needs a shared atomic limit and deadlines. Retry only retry-safe operations, honoring provider retry guidance; a timeout after a side effect is not evidence that nothing happened. See Tenacity retry predicates. …
How to use it
Reference it in AdaL, Claude Code, Cursor or any coding agent — nothing to install:
@skills sickn33/llm-app-patterns--979570