AgencyBench

Long-horizon agent benchmark: 32 scenarios, 138 tasks, ~1M tokens and ~90 tool calls; Docker sandbox and rubric-based + LLM judges.

Stars

87

Adoption surface

complex

Autonomy

headless

Recovery

none

License

✅ open-source

Category

Evaluation and benchmarking harnesses

Repository ↗ Example: AgencyBench leaderboard ↗

Related in Evaluation and benchmarking harnesses

• Agent Lightning
• SWE-bench
• AgentBench
• inspect_ai
• WebArena
• WebVoyager