Anti-Patterns: Workflow

Cramming an entire complex task into a single monolithic prompt, expecting the LLM to handle everything at once.

Agent enters an unbounded execution cycle — retrying failed actions, oscillating between states, or generating plans it never completes.

Using a multi-agent system when a simple prompt chain or single agent would suffice — adding architectural complexity without proportional benefit.

A supervisor agent delegates to sub-agents without verifying their output, trusting results uncritically.