-
Accept benchmark observation explicit-edit-2026-09-14T19-36-29.863Z
-
Accept benchmark observation explicit-edit-2026-09-14T16-39-38.050Z
-
Add a leaderboard by model to the dataset
-
Drop the trust field and compare runs by their rules
-
Declare one runner and one benchmark version in every bundle
-
Identify a run by its rules and task set, not by a commit
-
Name every harness by its published family
-
Name the accepted IDE observation by its harness family
-
Accept benchmark observation explicit-edit-2026-09-14T09-12-56.814Z
-
Remove the mislabeled pi-agent-ide 0.5.0 observation
-
Accept benchmark observation explicit-edit-2026-09-14T08-10-56.494Z
-
Name harnesses by family in the accepted baseline
-
Bootstrap 1 benchmark source bundles
-
Accept benchmark observation hf-user-pr-accept-schema1-recipe-20260913
-
Bootstrap 3 benchmark source bundles
-
Correct Pi Agent IDE baseline version to 0.2.2
-
Migrate agent, harness, and runner identities
-
Accept benchmark observation public-ide-0-3-0-luna-low-upload-smoke-3-20260911
-
Publish explicit-edit v1 baseline corpus
-
Rename public explicit-edit contract to v1
-
Publish homogeneous explicit-edit baseline corpus
-
Remove split baseline shards
-
Publish retained five-model benchmark corpus
-
Replace test smoke with retained benchmark corpus
-
Publish 1 accepted benchmark observations
