Evaluation
Measuring agents in a way that can decide something.
2026
A browser for agents needs more than one score
A Rust browser engine built for agents, compared with Chrome across 1,024 sites, and why text, links, pixels and scripts each need their own result.
Agent systemsEvaluationPlatforms
Agent memory that remembers the current answer
Building a multi-tenant memory service where identity bounds access, later facts supersede earlier ones, and tests grade the final answer rather than retrieval.
SelectedMemoryEvaluationSecurity
Correcting the grader without rewriting the run
A coding-agent evaluation whose first grader was wrong in both directions, and how original records, corrected verdicts and repairs stayed separate results.
SelectedEvaluationPlatformsVerification
Did your agent experiment test what you think it tested?
Wrong models, arms that never switched on, shifting workloads and features that ran without helping: checks that show an agent comparison contains the change.
SelectedAgent systemsEvaluationVerification
Grading agents without trusting the simulator
An evaluation lab where the oracle never calls simulator code, grades in ten gates, tests itself with mutations and can answer that it doesn't know.
SelectedAgent systemsEvaluationVerification
Every draft passed the judges
A pipeline of model judges passed every draft. What those scores measured, how a reader’s preference differs, and why taste and facts need separate evidence.
WritingAgent systemsEvaluationCraft
Two ways a tool description goes stale
Why a discovered tool definition is not permission to call it: re-checking catalog, policy and capability at call time, and deriving every format from one source.
SelectedAgent systemsSecurityInterfacesEvaluation
When the evaluator is the broken part
Sixteen generated apps passed their functional checks while the benchmark recorded two passes. Separating what was built, how it was built and what the observer could prove explains the gap.
EvaluationVerification
Whose mistake did the coding-agent benchmark catch?
Reference and empty-patch controls, test-file collisions and environment faults: how to adapt a coding-agent benchmark and tell an agent's failure from the harness's.
SelectedAgent systemsEvaluationVerification