build-with-agent-runtime

You can automate agent genome creation and benchmarking by generating genomes using AgentProfile or AgentSurfaces, then evaluating performance on benchmarks. This Skill streamlines the entire workflow from creation to evaluation.

An agent genome defines the core configuration of an AI agent. You need to optimize it through a gated loop when you want to systematically improve agent performance and certify wins on a frozen holdout dataset to prevent overfitting.

Yes, executing agent genome benchmarking and optimization requires both @tangle-network/agent-runtime and @tangle-network/agent-eval dependencies. These packages provide the necessary environment to run and evaluate agent performance.

The best way to optimize AI agents is through a gated optimization loop that certifies wins on a frozen holdout. This approach ensures performance improvements are validated against unseen data, preventing overfitting during agent genome optimization.

Yes, if you are developing an AI agent for a specific task, this workflow supports creating the genome, running it on a benchmark to evaluate performance, and optimizing it to observe how well it executes the target task.