Build LLM workflows with typed clients, schema validation, and versioned prompts.
npx skills add https://github.com/lciacci/tessera --skill llm-patterns-lciacci
Please help me install this Agent Skill. Skill: llm-patterns Source: https://github.com/lciacci/tessera/tree/main/skills/llm-patterns Command: npx skills add https://github.com/lciacci/tessera --skill llm-patterns-lciacci
This Skill helps developers build dependable AI-first applications without putting business logic, validation, testing, and operational safeguards entirely inside prompts or live LLM calls.
Use the llm-patterns skill to design a typed LLM client, versioned prompt templates, schema validation, deterministic tests, and a nightly evaluation workflow for my application.
High-intent search queries and answers about installing and using this skill.
Validate LLM outputs by structuring model clients with schema validation and typed responses. This separates LLM reasoning from application plumbing, ensuring your LLM outputs conform to expected data structures before processing.
Test prompt engineering workflows deterministically by using mocks and fixtures for LLM responses. This allows reliable testing of classification and extraction logic without relying on live model calls or unpredictable network latency.
Track token usage, model selection, latency, and estimated costs by implementing operational monitoring within your LLM application workflows. This enables continuous visibility into API consumption and operational expenses.
Yes, prompt management patterns allow you to organize reusable prompt templates and version prompt changes separately. This isolates LLM reasoning iterations from your core application logic and plumbing.
Measure LLM model accuracy by running scheduled evaluation suites. These nightly workflows test live model responses against expected outcomes to continuously evaluate production-quality accuracy for tasks like summarization and decision-making.
Design fallbacks for LLM API failures by building failure fallback mechanisms into your application workflows. This ensures your AI-first application maintains dependability even when live LLM calls experience errors or timeouts.
ViperJuiceGenerate typed DTOs and client wrappers for Python and TypeScript projects.
fjpulidopAnalyze per-agent execution telemetry from CLAUDE CLI JSONL logs and metadata.
Kingly-AgencyDesigns complex prompts for production LLM applications with few-shot learning and chain-of-thought reasoning.
alexpopovCreate, update, and search Notion pages and databases via REST API calls.
