v4.38.1

What's Changed

• ci(release): gate finalize before bump and use GH_MODELS_TOKEN by @christso in #1394
• fix(cli): make required_version advisory, warn on eval failure instead of prompting/exiting by @christso in #1396

Full Changelog: v4.38.0...v4.38.1

v4.38.1-next.1

What's Changed

• ci(release): gate finalize before bump and use GH_MODELS_TOKEN by @christso in #1394
• fix(cli): make required_version advisory, warn on eval failure instead of prompting/exiting by @christso in #1396

Full Changelog: v4.38.0...v4.38.1-next.1

v4.38.0

What's Changed

• feat(evaluation): final-answer output with trace artifacts by @christso in #1364
• feat(results): support configured storage branch by @christso in #1365
• fix(pi-cli): set AZURE_OPENAI_BASE_URL for full URLs, AZURE_OPENAI_RESOURCE_NAME for bare names by @christso in #1367
• feat(results): periodic WIP checkpoints for in-progress eval runs by @christso in #1369
• docs(results): document WIP checkpoints by @christso in #1371
• feat(dashboard): render structured transcript artifacts by @christso in #1370
• docs(ops): remove Agent Mail coordination instructions by @christso in #1372
• fix(results): harden static report publishing by @christso in #1373
• feat(replay): record and replay target outputs by @christso in #1374
• docs(agents): remove local override file guidance by @christso in #1378
• docs(trace): specify trace envelope contract by @christso in #1379
• feat(copilot): support custom provider config by @christso in #1382
• feat(trace): write trace envelope sidecars by @christso in #1376
• feat(replay): support trace envelope replay sources by @christso in #1381
• feat(dashboard): improve sidebar navigation by @christso in #1383
• feat(dashboard): move project context to top bar by @christso in #1384
• refactor(dashboard): make project context read-only by @christso in #1386
• refactor(dashboard): move run context into breadcrumbs by @christso in #1387
• feat(workspace): split repo provenance from acquisition by @christso in #1390
• ci(release): add contract eval gate before latest publish by @christso in #1393

Full Changelog: v4.35.1...v4.38.0

v4.38.0-next.1

What's Changed

• feat(dashboard): move project context to top bar by @christso in #1384
• refactor(dashboard): make project context read-only by @christso in #1386
• refactor(dashboard): move run context into breadcrumbs by @christso in #1387
• feat(workspace): split repo provenance from acquisition by @christso in #1390

Full Changelog: v4.37.0-next.1...v4.38.0-next.1

v4.37.0-next.1

What's Changed

• docs(agents): remove local override file guidance by @christso in #1378
• docs(trace): specify trace envelope contract by @christso in #1379
• feat(copilot): support custom provider config by @christso in #1382
• feat(trace): write trace envelope sidecars by @christso in #1376
• feat(replay): support trace envelope replay sources by @christso in #1381
• feat(dashboard): improve sidebar navigation by @christso in #1383

Full Changelog: v4.36.0-next.1...v4.37.0-next.1

v4.36.0-next.1

What's Changed

• feat(evaluation): final-answer output with trace artifacts by @christso in #1364
• feat(results): support configured storage branch by @christso in #1365
• fix(pi-cli): set AZURE_OPENAI_BASE_URL for full URLs, AZURE_OPENAI_RESOURCE_NAME for bare names by @christso in #1367
• feat(results): periodic WIP checkpoints for in-progress eval runs by @christso in #1369
• docs(results): document WIP checkpoints by @christso in #1371
• feat(dashboard): render structured transcript artifacts by @christso in #1370
• docs(ops): remove Agent Mail coordination instructions by @christso in #1372
• fix(results): harden static report publishing by @christso in #1373
• feat(replay): record and replay target outputs by @christso in #1374

Full Changelog: v4.35.1...v4.36.0-next.1

v4.35.1

What's Changed

• fix(eval): accept structured code grader content by @christso in #1360
• fix(targets): add Pi SDK OpenAI target by @christso in #1361

Full Changelog: v4.35.0...v4.35.1

v4.35.1-next.1

What's Changed

• fix(eval): accept structured code grader content by @christso in #1360
• fix(targets): add Pi SDK OpenAI target by @christso in #1361

Full Changelog: v4.35.0...v4.35.1-next.1

v4.35.0

What's Changed

• feat(scripts): deploy example eval results repos by @christso in #1267
• fix(docker): run compose service as host user by @christso in #1268
• [codex] Ignore Gas Town runtime files by @christso in #1269
• feat(docker): add dashboard deployment script by @christso in #1271
• chore(dashboard): rename app directory by @christso in #1273
• docs(agents): document AO-first workflow by @christso in #1275
• docs(agents): align AO task graph guidance by @christso in #1276
• feat(config): split AgentV home and data directories by @christso in #1278
• feat(phoenix): add AgentV eval adapter by @christso in #1279
• test(cli): simplify churny local coverage by @christso in #1289
• docs(orchestration): use br-first Beads workflow by @christso in #1288
• docs(agents): clarify regression-focused tests by @christso in #1290
• fix(dashboard): support white-label app branding by @christso in #1291
• chore(beads): track shared bead state by @christso in #1292
• docs(agents): remove redundant bead guidance by @christso in #1293
• feat(codex): support reasoning effort targets config by @christso in #1294
• fix(codex): use stream_log for target logging by @christso in #1295
• feat(results): support global per-project results config by @christso in #1296
• fix(config): complete layered home config migration by @christso in #1299
• fix(targets): preserve azure deployment base urls by @christso in #1301
• docs(pi): document thinking level config by @christso in #1302
• feat(results): delete local run workspaces by @christso in #1300
• feat(dashboard): add project results sync UX by @christso in #1303
• docs(evals): document benchmark provenance patterns by @christso in #1304
• fix(core): allow templated use_target validation by @christso in #1305
• fix(cache): honor configured response cache paths by @christso in #1307
• feat(showcase): add replay-first trace evaluation fixtures by @christso in #1308
• feat(evals): preserve rubric criterion operators by @christso in #1309
• docs(dashboard): document project results sync workflow by @christso in #1306
• fix(dashboard): use registry project display names by @christso in #1310
• fix(dashboard): reconcile remote run counts by @christso in #1311
• fix(dashboard): preserve remote run detail context by @christso in #1312
• feat(dashboard): clarify remote sync outcome by @christso in #1313
• feat(results): publish selected local runs by @christso in #1314
• fix(results): remove selected run publish workflow by @christso in #1315
• docs(evals): plan AgentV eval authoring extensibility by @christso in #1316
• docs(integrations): document Phoenix adapter by @christso in #1317
• fix(dashboard): clarify remote run actions by @christso in #1318
• fix(dashboard): make run tables mobile scrollable by @christso in #1319
• style(brand): present AgentV wordmark by @christso in #1320
• fix(results): separate execution errors from quality failures by @christso in #1321
• chore(beads): record AgentV demo gap progress by @christso in #1322
• style(brand): uppercase visible AGENTV wordmark by @christso in #1323
• fix(web): dedupe landing wordmark by @christso in #1324
• feat(dashboard): filter score distribution by @christso in #1327
• fix(dashboard): remove redundant quality label prefixes by @christso in #1328
• docs(agents): clarify beads ownership for research by @christso in #1333
• feat(results): materialize per-test task bundles by @christso in #1330
• docs(agents): remove local orchestration guidance by @christso in #1334
• fix(results): harden remote sync dogfood by @christso in #1332
• feat(trace): add normalized trajectory contract by @christso in #1331
• feat(cli): rerun captured task bundles by @christso in #1335
• feat(cli): simplify eval output surface by @christso in #1336
• fix(results): supersede stale remote sync errors by @christso in #1338
• docs(agents): externalize task tracker guidance by @christso in #1339
• chore(tracking): remove repo-local beads state by @christso in #1340
• fix(dashboard): compact run labels by @christso in #1337
• feat(core): support structured llm-grader context by @christso in #1342
• fix(dashboard): settle remote sync button state by @christso in #1343
• fix(results): retry interrupted direct result pushes by @christso in #1344
• docs(results): clarify remote sync CLI contract by @christso in #1345
• docs(run): document generated task bundles by @christso in #1350
• fix(results): ignore unrelated files during remote sync by @christso in #1347
• fix(cli): remove eval benchmark json flag by @christso in #1348
• fix(dashboard): make targets table scroll on mobile by @christso in #1346
• fix(cli): remove eval export option by @christso in #1349
• fix(dashboard): prevent mobile experiment table clipping by @christso in #1352
• docs(adr): keep Phoenix observability out of core by @christso in #1353
• test(core): tolerate grader timing jitter by @christso in #1355
• feat(config): hard-break dashboard project schema by @christso in #1356
• feat(config): use succinct repo_url project remotes by @christso in #1357
• fix(dashboard): sort runs by timestamp by @christso in #1359

Full Changelog: v4.31.3...v4.35.0

v4.35.0-next.1

What's Changed

• test(core): tolerate grader timing jitter by @christso in #1355
• feat(config): hard-break dashboard project schema by @christso in #1356
• feat(config): use succinct repo_url project remotes by @christso in #1357
• fix(dashboard): sort runs by timestamp by @christso in #1359

Full Changelog: v4.34.1-next.1...v4.35.0-next.1