15 prompts
Fuzz your API endpoints with random but valid inputs. Find edge cases, crashes, and 500 errors that manual testing misses.
Build an AI agent that reasons and acts using tools. ReAct loop with function calling, error recovery, and conversation memory.
Build type-safe server actions for Next.js forms. Validation, error handling, optimistic updates, and revalidation — the modern mutation pattern.
Set up a complete observability stack. Structured logging, Prometheus metrics, distributed tracing with OpenTelemetry — correlated with trace IDs.
Reliably extract structured JSON from LLM responses. Schema validation, retry on malformed output, and fallback parsing strategies.
Build a semantic code search engine using embeddings. Search by intent ('function that validates email') rather than exact text matches.
Build a fast local-first backend using Bun's built-in SQLite. CRUD API, migrations, full-text search, and WAL mode for concurrent reads.
Build a Retrieval-Augmented Generation pipeline with hybrid search (vector + keyword) and a reranking step for higher precision answers.
Build a prompt template engine that supports variables, conditionals, loops, and includes. Like Handlebars but for LLM prompts.
Build a full tRPC v11 setup with procedures, middleware, subscriptions, and React Query integration. Full-stack TypeScript type safety.
Track LLM API costs across multiple providers. Token counting, cost calculation, budget alerts, and per-feature usage attribution.
Build a GitHub PR review bot that uses LLMs to analyze code changes, detect issues, and post review comments automatically.
Build an edge function that routes requests by geography, applies rate limiting, and transforms responses — all at the edge for minimal latency.
Build a REST API with Deno 2 using Web Standards APIs. Secure by default with explicit permissions, built-in testing, and zero config.
Implement streaming LLM responses with Server-Sent Events. Token-by-token output from API to browser with proper error handling and abort.