Commit Graph

  • dfeae09526 fix(ce-code-review): stop the reviewer template from embedding the diff twice (#1512) Trevin Chow 2026-08-22 08:57:36 -07:00
  • c85d44a0a0 fix(ce-code-review): stop the reviewer template from embedding the diff twice tmchow/issue-1509-eval Trevin Chow 2026-08-22 08:40:39 -07:00
  • f1a80f216f chore: release main (#1434) compound-engineering-v3.23.0 github-actions[bot] 2026-08-22 00:54:30 -07:00
  • f7efe88675 docs(agent-plugins): attribute the 8000-byte skill cap to its real owner (#1511) Trevin Chow 2026-08-22 00:32:13 -07:00
  • a823da2750 Address PR review feedback (#1511) tmchow/fix-skill-md-8kb Trevin Chow 2026-08-21 23:26:29 -07:00
  • a08ed61e41 docs(agent-plugins): drop the accepted-risk paragraph Trevin Chow 2026-08-21 23:20:19 -07:00
  • fa165d7456 docs(agent-plugins): record the combined-budget gap as an accepted risk Trevin Chow 2026-08-21 23:20:02 -07:00
  • d7f24ee36c Address PR review feedback (#1511) Trevin Chow 2026-08-21 23:18:22 -07:00
  • 1e57deb90f docs(agent-plugins): attribute the 8000-byte skill cap to its real owner Trevin Chow 2026-08-21 23:08:25 -07:00
  • 28ac832285 fix(cross-model): accept provider-qualified codex model ids in cross_model_model (#1501) Dennis Traub 2026-08-22 07:25:26 +02:00
  • 9107007954 test(cross-model): give the subprocess-heavy route suites a 30s per-test ceiling (#1510) Trevin Chow 2026-08-21 15:59:16 -07:00
  • 72de1d51f7 test(cross-model): give the subprocess-heavy route suites a 30s per-test ceiling tmchow/cross-model-routes-test-timeout Trevin Chow 2026-08-21 15:45:57 -07:00
  • bcd86d0449 fix(skills): load ce-plan and ce-work phases from references under the 8KB cap (#1508) Trevin Chow 2026-08-21 15:36:14 -07:00
  • f14da4a2f7 fix(ce-work): a read made before a phase does not satisfy its acting-point read tmchow/eval-issue-1482-skill Trevin Chow 2026-08-21 15:13:04 -07:00
  • 6dda2038b1 fix(ce-optimize): keep multi-objective wins and screen expensive runs (#1506) Trevin Chow 2026-08-21 15:12:24 -07:00
  • 0ed254a6ac fix(ce-plan): a read made before a phase does not satisfy its acting-point read Trevin Chow 2026-08-21 15:02:11 -07:00
  • bcb92cae5b Address PR review feedback (#1506) tmchow/audit-ce-optimize-skill Trevin Chow 2026-08-21 14:57:09 -07:00
  • 3efb999d5e docs(skill-design): record the phase-loaded kernel plan, its evaluation, and two learnings Trevin Chow 2026-08-21 14:51:56 -07:00
  • 537edade8d fix(skills): load ce-plan and ce-work phase mechanics from references under the 8,000-byte contract Trevin Chow 2026-08-21 14:51:49 -07:00
  • 545c0736ff Address PR review feedback (#1506) Trevin Chow 2026-08-21 14:50:36 -07:00
  • b1f78dbd0b Address PR review feedback (#1506) Trevin Chow 2026-08-21 14:42:33 -07:00
  • fe37662745 Address PR review feedback (#1506) Trevin Chow 2026-08-21 14:37:03 -07:00
  • 572faf494a docs(skill-design): qualify the authoring standard with two 2026-08-21 eval learnings Trevin Chow 2026-08-21 14:35:57 -07:00
  • 35c8bc4622 Address PR review feedback (#1506) Trevin Chow 2026-08-21 14:32:22 -07:00
  • dad3777bd0 Address PR review feedback (#1506) Trevin Chow 2026-08-21 14:27:31 -07:00
  • 3f76802bbd Address PR review feedback (#1506) Trevin Chow 2026-08-21 14:20:38 -07:00
  • 0259625a23 Address PR review feedback (#1506) Trevin Chow 2026-08-21 14:13:17 -07:00
  • 29c85cdd0c Address PR review feedback (#1506) Trevin Chow 2026-08-21 14:08:23 -07:00
  • a9f0f22f88 Address PR review feedback (#1506) Trevin Chow 2026-08-21 14:03:02 -07:00
  • 89e3c8b760 Address PR review feedback (#1506) Trevin Chow 2026-08-21 13:56:15 -07:00
  • 08d91194eb Address PR review feedback (#1506) Trevin Chow 2026-08-21 13:47:12 -07:00
  • 99e33f186b Address PR review feedback (#1506) Trevin Chow 2026-08-21 13:42:36 -07:00
  • 3cd5590263 Address PR review feedback (#1506) Trevin Chow 2026-08-21 13:35:16 -07:00
  • 0e2e2d7666 Address PR review feedback (#1506) Trevin Chow 2026-08-21 13:30:20 -07:00
  • a0ddecbb24 Address PR review feedback (#1506) Trevin Chow 2026-08-21 12:16:11 -07:00
  • 5c82583d00 Address PR review feedback (#1506) Trevin Chow 2026-08-21 11:59:48 -07:00
  • 653f6c7d65 Address PR review feedback (#1506) Trevin Chow 2026-08-21 11:49:20 -07:00
  • 71fa984127 fix(ce-optimize): default judge floor, rank relatively, and distinguish censor (#1506) Trevin Chow 2026-08-21 11:29:35 -07:00
  • e3f3b211b0 fix(ce-optimize): accept nested spec and fail closed on gates (#1506) Trevin Chow 2026-08-21 11:13:14 -07:00
  • ae745a19bb fix(ce-optimize): honor declared bars and stop false futility (#1506) Trevin Chow 2026-08-21 10:58:12 -07:00
  • 4d4d8cece9 fix(ce-optimize): keep primary required and honor spec bounds (#1506) Trevin Chow 2026-08-21 10:43:20 -07:00
  • c8dc8a5a02 fix(ce-optimize): keep multi-objective wins and screen expensive runs Trevin Chow 2026-08-21 10:28:39 -07:00
  • 66ccf579f8 fix(skills): remove obsolete dispatch context hook (#1499) Trevin Chow 2026-08-20 23:30:13 -07:00
  • fb56a64a59 fix(skills): remove obsolete dispatch context hook tmchow/issue-1481-skills-eval Trevin Chow 2026-08-20 23:07:47 -07:00
  • dedf301348 fix: honor active model-config keys in skill runs (surface + salience) (#1489) Kevin Old 2026-08-20 22:28:11 -05:00
  • 3c04954458 fix(cross-model): elevate peer start when host sandbox blocks provider network (#1496) Trevin Chow 2026-08-20 17:59:09 -07:00
  • bb0203f597 fix(review): keep scoped standards host-owned (#1496) tmchow/codex-sandbox-cross-model-start-b8c6 Trevin Chow 2026-08-20 17:50:43 -07:00
  • 736a5012dd fix(review): separate trusted peer constraints (#1496) Trevin Chow 2026-08-20 17:09:24 -07:00
  • bac0f385da fix(review): preserve sandboxed peer context (#1496) Trevin Chow 2026-08-20 16:47:46 -07:00
  • f418e86879 fix(ce-work): preserve fail-closed skill loading (#1496) Trevin Chow 2026-08-20 16:39:43 -07:00
  • e1f200aa59 fix(ce-work): fall back without sandbox escalation (#1496) Trevin Chow 2026-08-20 16:31:49 -07:00
  • 9c6b8a7344 fix(skills): defer POV auth proof to dispatch (#1496) Trevin Chow 2026-08-20 13:21:15 -07:00
  • 67aed8fa2f fix(skills): preserve auth fallback coverage (#1496) Trevin Chow 2026-08-20 13:11:08 -07:00
  • 9367a705f1 fix(skills): classify auth after provider dispatch (#1496) Trevin Chow 2026-08-20 12:48:07 -07:00
  • 4572c3b6f4 fix(skills): prove elevation auth at dispatch (#1496) Trevin Chow 2026-08-20 12:39:51 -07:00
  • c198164745 fix(skills): narrow sandbox elevation scope (#1496) Trevin Chow 2026-08-20 12:28:48 -07:00
  • 981b8220db fix(ce-work): close dispatch denial race (#1496) Trevin Chow 2026-08-20 12:00:08 -07:00
  • 60939f1b99 fix(skills): restore cross-model dispatch from Codex sandboxes tmchow/fix-sandbox-env-skill Trevin Chow 2026-08-20 11:34:52 -07:00
  • 12d3d8c3df fix(skills): unify needs-human decision handoff (#1492) Trevin Chow 2026-08-20 10:13:57 -07:00
  • 9f55d68a96 fix(ce-resolve-pr-feedback): keep replies out of pending reviews (#1491) Trevin Chow 2026-08-20 10:13:57 -07:00
  • a46567427c fix: synchronize heartbeat startup tmchow/fix-needs-human-handoff Trevin Chow 2026-08-20 09:51:46 -07:00
  • 874cb9af79 fix: make heartbeat teardown race-free Trevin Chow 2026-08-20 09:45:07 -07:00
  • 119d4246bb fix: reap peer heartbeat process groups Trevin Chow 2026-08-20 09:38:51 -07:00
  • 2caf0ced74 fix: bind decision thread URLs to sources Trevin Chow 2026-08-20 08:25:06 -07:00
  • d538a1f031 fix: retain reactivated currency on head moves Trevin Chow 2026-08-20 08:12:13 -07:00
  • d42a2f3cc8 fix: reject mixed decision answer actions Trevin Chow 2026-08-20 08:08:43 -07:00
  • 07e03531e1 fix(skills): unify needs-human decision handoff Trevin Chow 2026-08-20 08:00:07 -07:00
  • 897ed00278 fix: inherit snapshot test timeout tmchow/simplify-needs-human-ledger Trevin Chow 2026-08-20 03:01:12 -07:00
  • 2bbfdee02a fix: let decisions adopt dispatched sources Trevin Chow 2026-08-20 02:40:33 -07:00
  • 687725ab97 fix: report only completed PR actions Trevin Chow 2026-08-20 02:26:25 -07:00
  • 7582081303 fix: align terminal currency recovery Trevin Chow 2026-08-20 02:18:58 -07:00
  • 9f62977360 fix: render read-only answer transitions Trevin Chow 2026-08-20 02:12:19 -07:00
  • df53a2dc46 test: use shared snapshot timeout Trevin Chow 2026-08-20 02:00:33 -07:00
  • db0b134f88 fix: retain currency decision authorization Trevin Chow 2026-08-20 01:56:29 -07:00
  • c10a523fbb fix: make decision answer marks self-contained Trevin Chow 2026-08-20 01:48:38 -07:00
  • 3b9ba18cfd fix: preserve exact legacy answer evidence Trevin Chow 2026-08-20 01:37:19 -07:00
  • 8642eb8a6b fix: reactivate removed decision sources Trevin Chow 2026-08-20 01:23:22 -07:00
  • 4cc4e3270a fix: reactivate answered decision sources Trevin Chow 2026-08-20 01:05:49 -07:00
  • 7abb1bd7c9 fix: publish ambiguous currency decisions Trevin Chow 2026-08-20 00:53:06 -07:00
  • aec7442b4f fix: preserve currency decision authority Trevin Chow 2026-08-20 00:39:11 -07:00
  • acd4feae87 fix: make decision coverage head-bound Trevin Chow 2026-08-20 00:17:55 -07:00
  • e5f8e04ff0 fix: scope currency decisions to active evidence Trevin Chow 2026-08-20 00:00:06 -07:00
  • 6c4349e826 refactor: unify needs-human decision state Trevin Chow 2026-08-19 23:44:28 -07:00
  • 33f66b5901 test: classify feature-only skill evals Trevin Chow 2026-08-19 22:18:19 -07:00
  • 3651b3f903 fix: reject misplaced currency decisions Trevin Chow 2026-08-19 22:18:15 -07:00
  • fbbf2efab5 fix: reject ignored decision payloads Trevin Chow 2026-08-19 21:58:19 -07:00
  • dbd0d258db fix: identify legacy status reruns Trevin Chow 2026-08-19 21:45:17 -07:00
  • 2d454f7f7d fix: retain pending currency decisions Trevin Chow 2026-08-19 21:29:51 -07:00
  • 7e4709be66 fix: preserve currency decision history Trevin Chow 2026-08-19 21:16:26 -07:00
  • 0c1e3d7e66 fix: reopen changed check decisions Trevin Chow 2026-08-19 20:53:22 -07:00
  • 33dabba8fe fix: preserve unrelated semantic decisions Trevin Chow 2026-08-19 20:48:57 -07:00
  • 567a43db9c fix: complete semantic decision recovery Trevin Chow 2026-08-19 20:33:43 -07:00
  • a89d5e147d fix: preserve decision ownership across source changes Trevin Chow 2026-08-19 19:40:30 -07:00
  • 0e3aadac6f fix: enforce the canonical decision residual Trevin Chow 2026-08-19 19:07:47 -07:00
  • 034d68ad10 Address atomic residual review feedback (#1492) Trevin Chow 2026-08-19 18:40:48 -07:00
  • d3b89890ae Address shared residual review feedback (#1492) Trevin Chow 2026-08-19 18:23:31 -07:00
  • c95b9287d7 Address follow-up PR feedback (#1492) Trevin Chow 2026-08-19 18:00:26 -07:00
  • 45d17d5509 Address PR review feedback (#1492) Trevin Chow 2026-08-19 17:51:45 -07:00
  • 1ab9c827d6 fix(skills): surface needs-human handoff decisions Trevin Chow 2026-08-19 17:40:45 -07:00
  • 93b4261b52 fix: reconcile reply completion across modes tmchow/fix-reply-to-pr Trevin Chow 2026-08-19 20:43:09 -07:00