A multi-session coding agent benchmark measuring requirement retention, architectural boundary integrity, and token efficiency across context resets.
python benchmark ai-agents multi-session llm-evaluation coding-agents token-efficiency context-compaction software-engineering-benchmark
-
Updated
Nov 21, 2026 - Python