agent-evaluation

Agent evaluation requires outcome-focused approaches that account for non-determinism and multiple valid paths. Multi-dimensional rubrics capture various quality aspects: factual accuracy, completeness, citation accuracy, source quality, and tool efficiency. LLM-as-judge provides scalable evaluation while human evaluation catches edge cases.

The key insight is that agents may find alternative paths to goals—the evaluation should judge whether they achieve right outcomes while following reasonable processes.

Performance Drivers: The 95% Finding

Research on the BrowseComp evaluation (which tests browsing agents' ability to locate hard-to-find information) found that three factors explain 95% of performance variance:

| Factor | Variance Explained | Implication |

|--------|-------------------|-------------|

| Token usage | 80% | More tokens = better performance |

| Number of tool calls | ~10% | More exploration helps |

| Model choice | ~5% | Better models multiply efficiency |

Implications for Claude Code development:

• Token budgets matter: Evaluate with realistic token constraints
• Model upgrades beat token increases: Upgrading models provides larger gains than increasing token budgets
• Multi-agent validation: Validates architectures that distribute work across subagents with separate context windows

L2 Content (loaded when methodology details needed):

• See: [references/methodologies.md](./references/methodologies.md)

- Evaluation Challenges

- Evaluation Rubric Design

- Evaluation Methodologies

- Test Set Design

- Context Engineering Evaluation

L3 Content (loaded when advanced techniques or examples needed):

• See: [references/advanced.md](./references/advanced.md)

- Advanced Evaluation: LLM-as-Judge

- Evaluation Metrics Reference

- Bias Mitigation Techniques

- LLM-as-Judge Implementation Patterns

- Metric Selection Guide

- Practical Examples