v4.42.4

What's Changed

• fix(evals): require self-eval guidance reads by @christso in #1447
• fix(results): root results branch at a stable deterministic genesis by @christso in #1448

Full Changelog: v4.42.3...v4.42.4

v4.42.4-next.1

What's Changed

• fix(evals): require self-eval guidance reads by @christso in #1447
• fix(results): root results branch at a stable deterministic genesis by @christso in #1448

Full Changelog: v4.42.3...v4.42.4-next.1

v4.42.3

What's Changed

• fix(artifacts): write trace sidecars to canonical path by @christso in #1446

Full Changelog: v4.42.2...v4.42.3

v4.42.3-next.1

What's Changed

• fix(artifacts): write trace sidecars to canonical path by @christso in #1446

Full Changelog: v4.42.2...v4.42.3-next.1

v4.42.2

What's Changed

• fix(dashboard): show eval suite labels by @christso in #1397
• docs(workspace): document Harbor execution boundary by @christso in #1398
• fix(results): handle replay row schema aliases by @christso in #1400
• refactor(observability): move backend presets out of core by @christso in #1399
• feat(results): publish runs to results branch by @christso in #1403
• feat(trace): settle canonical trace projection by @christso in #1401
• docs(evals): clarify grader wire format by @christso in #1404
• feat(copilot): flat pi-style provider config, SDK args, ACP custom providers by @christso in #1402
• fix(trace): use agentv.trace.v1 for canonical trace artifacts by @christso in #1405
• docs(adr): decide SDK eval authoring boundary by @christso in #1411
• feat(sdk): let evaluate write canonical run artifacts by @christso in #1410
• chore(sdk): remove deprecated aliases and scaffold drift by @christso in #1407
• docs(sdk-dx): align eval authoring surfaces by @christso in #1409
• feat(contract): add code-grader stdin payload contract eval by @christso in #1413
• fix(copilot): trust subprocess cwd for relative path resolution in args by @christso in #1412
• fix(contract): avoid release eval content filter by @christso in #1414
• docs(export): define Opik post-run boundary by @christso in #1406
• feat(sdk): add minimal python authoring helpers by @christso in #1408
• refactor(results): rename WIP checkpoint branch namespace by @christso in #1415
• docs(targets): clarify Pi CLI custom endpoint behavior by @christso in #1416
• fix(copilot): use api_format for custom provider wire format by @christso in #1417
• ci: update GitHub Actions for Node 24 runners by @christso in #1418
• docs(strategy): define repo-native product boundary by @christso in #1420
• docs(roadmap): clarify Harbor runner direction by @christso in #1428
• docs(agents): split root guide into linked index by @christso in #1429
• feat(trace): prepare trace artifact grading by @christso in #1419
• feat(cli): bundle eval suites for portable reruns by @christso in #1421
• fix(dashboard): enforce configured run source contract by @christso in #1426
• feat(sdk): replace @agentv/eval with @agentv/sdk by @christso in #1427
• fix(release): publish scoped packages as public by @christso in #1431
• fix(skill-trigger): case-insensitive read tool name and support input.path key by @christso in #1434
• feat(sdk): add yaml-aligned eval authoring helpers by @christso in #1430
• feat(dashboard): add project removal action by @christso in #1436
• refactor(evals): replace stale self skill fixtures by @christso in #1435
• fix(release): keep finalize versions on main by @christso in #1437
• feat(sdk): add grader helper catalog by @christso in #1438
• fix(results): flatten on-branch results layout and tolerate missing branch by @christso in #1439
• fix(dashboard): replace source filter with per-run remote indicator by @christso in #1441
• docs(sdk): update public guidance for sdk migration by @christso in #1440
• feat(export): add stable projection identity by @christso in #1442
• feat(dashboard): add canonical result table by @christso in #1443
• feat(artifacts): normalize transcript sidecar contract by @christso in #1445

Full Changelog: v4.38.1...v4.42.2

v4.42.2-next.1

What's Changed

• docs(sdk): update public guidance for sdk migration by @christso in #1440
• feat(export): add stable projection identity by @christso in #1442
• feat(dashboard): add canonical result table by @christso in #1443
• feat(artifacts): normalize transcript sidecar contract by @christso in #1445

Full Changelog: v4.42.1-next.1...v4.42.2-next.1

v4.42.1-next.1

What's Changed

• refactor(evals): replace stale self skill fixtures by @christso in #1435
• fix(release): keep finalize versions on main by @christso in #1437
• feat(sdk): add grader helper catalog by @christso in #1438
• fix(results): flatten on-branch results layout and tolerate missing branch by @christso in #1439
• fix(dashboard): replace source filter with per-run remote indicator by @christso in #1441

Full Changelog: v4.42.0-next.1...v4.42.1-next.1

v4.42.0-next.1

What's Changed

• feat(dashboard): add project removal action by @christso in #1436

Full Changelog: v4.41.6-next.1...v4.42.0-next.1

v4.41.6

What's Changed

• fix(dashboard): show eval suite labels by @christso in #1397
• docs(workspace): document Harbor execution boundary by @christso in #1398
• fix(results): handle replay row schema aliases by @christso in #1400
• refactor(observability): move backend presets out of core by @christso in #1399
• feat(results): publish runs to results branch by @christso in #1403
• feat(trace): settle canonical trace projection by @christso in #1401
• docs(evals): clarify grader wire format by @christso in #1404
• feat(copilot): flat pi-style provider config, SDK args, ACP custom providers by @christso in #1402
• fix(trace): use agentv.trace.v1 for canonical trace artifacts by @christso in #1405
• docs(adr): decide SDK eval authoring boundary by @christso in #1411
• feat(sdk): let evaluate write canonical run artifacts by @christso in #1410
• chore(sdk): remove deprecated aliases and scaffold drift by @christso in #1407
• docs(sdk-dx): align eval authoring surfaces by @christso in #1409
• feat(contract): add code-grader stdin payload contract eval by @christso in #1413
• fix(copilot): trust subprocess cwd for relative path resolution in args by @christso in #1412
• fix(contract): avoid release eval content filter by @christso in #1414
• docs(export): define Opik post-run boundary by @christso in #1406
• feat(sdk): add minimal python authoring helpers by @christso in #1408
• refactor(results): rename WIP checkpoint branch namespace by @christso in #1415
• docs(targets): clarify Pi CLI custom endpoint behavior by @christso in #1416
• fix(copilot): use api_format for custom provider wire format by @christso in #1417
• ci: update GitHub Actions for Node 24 runners by @christso in #1418
• docs(strategy): define repo-native product boundary by @christso in #1420
• docs(roadmap): clarify Harbor runner direction by @christso in #1428
• docs(agents): split root guide into linked index by @christso in #1429
• feat(trace): prepare trace artifact grading by @christso in #1419
• feat(cli): bundle eval suites for portable reruns by @christso in #1421
• fix(dashboard): enforce configured run source contract by @christso in #1426
• feat(sdk): replace @agentv/eval with @agentv/sdk by @christso in #1427
• fix(release): publish scoped packages as public by @christso in #1431
• fix(skill-trigger): case-insensitive read tool name and support input.path key by @christso in #1434
• feat(sdk): add yaml-aligned eval authoring helpers by @christso in #1430

Full Changelog: v4.38.1...v4.41.6

v4.41.6-next.1

What's Changed

• feat(sdk): add yaml-aligned eval authoring helpers by @christso in #1430

Full Changelog: v4.41.5-next.1...v4.41.6-next.1