You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Follow-up to #6. Cases TASK-INIT-enrich-ready and TASK-INIT-enrich-readonly were added to evals/mcp-tool-use/cases/task-initialization.ndjson and build under --dry-run, but no paid API run was authorized. The ready path was exercised once live on Claude Haiku 5.5 in an isolated client; the read-only failure path (outcome failed) has not been run. Run both cases per evals/README.md when an API key is available.
Activity