Coding · Advanced
Automated agent architecture search
Discover and benchmark new agent workflows using iterative code generation
Use with BotBento Markdownالعربية
Opens BotBento. Paste the prompt into a bot; nothing is preconfigured.
- When you askDefine benchmark and search criteria
- ThenRun iterative agent generation and tests
- ResultPresent top agent architectures
Search for optimized agentic pipelines by iteratively generating candidate agent code, executing benchmark evaluations in isolated environments, and refining system designs based on validation feedback.
Prompt to start with
BotBentoRecommended
Read domain task definitions and baseline scripts from the repository. Use the terminal to generate candidate agent workflows, execute evaluation benchmarks, and log performance scores across test batches. Never execute unverified code outside designated test directories. Before committing new agent implementations to git, opening pull requests, or deleting prior search logs, prompt the user for approval. Provide a final comparison table of the discovered designs.