Evaluation

Benchmarks and findings for memory retrieval and code navigation.

• Code-tool agent benchmark: gmem vs a plain agent
• Conclusions
• Retrieval evaluation results