Default Branch

9af474a70e · fix(skill-eval): record every Codex agent message in a conversation turn (#1822) · Updated 2026-10-02 05:14:12 +02:00

Branches

1a96d536c5 · fix(ce-debug): leave committing the structural fix to the handoff · Updated 2026-09-28 10:49:14 +02:00

22
3

6fb4ca9f40 · fix(test): read jobId from overrides in the ce-work harness · Updated 2026-09-28 10:42:30 +02:00

21
1

b5ad7a6653 · fix(test): tear down the helper's group on SIGHUP and SIGQUIT too · Updated 2026-09-28 10:25:19 +02:00

23
16

0f1f9531eb · fix(ce-resolve-pr-feedback): judge a defect in an earlier review fix by the same bound · Updated 2026-09-28 09:54:48 +02:00

28
3

8985e68861 · fix(ce-optimize): confirm the selection's own gain on a disjoint holdout · Updated 2026-09-28 05:33:48 +02:00

28
21

f8e606abc3 · fix(review): apply review findings · Updated 2026-09-28 00:49:21 +02:00

30
3

cf116f95ca · fix(cross-model): stop peer workers stalling on large failure logs · Updated 2026-09-28 00:03:01 +02:00

31
1

33c11e8dd7 · feat(site): add animated compound engineering explainer to the homepage · Updated 2026-09-25 20:48:46 +02:00

32
1

a87a7d3288 · fix(testing): keep boundary-call assertions valid in ce-work's test bar · Updated 2026-09-24 19:42:20 +02:00

38
2

44195c1bf6 · fix(ce-polish): install riffrec@^2.2.1 from npm instead of the git pin · Updated 2026-09-24 18:39:12 +02:00

37
1

6ea00fb596 · fix(ce-compound): drop frontend_react from the suggested component values · Updated 2026-09-23 19:13:51 +02:00

38
7

386f0a8684 · ci: re-run windows-native after a second hung Windows smoke step · Updated 2026-09-23 19:09:41 +02:00

38
7

ed3d05ae4a · fix(review): apply review findings · Updated 2026-09-23 18:14:40 +02:00

38
4

52411d6e0e · fix(resolve-pr-feedback): adjudicate judgment calls before escalating · Updated 2026-09-22 23:25:16 +02:00

39
1

f0020be7d9 · fix(models): target GPT-6 Sol and Luna for Codex peers and examples · Updated 2026-09-22 21:00:58 +02:00

41
1

32d52d2fe7 · fix(models): treat claude-opus-5 as rejecting sampling params · Updated 2026-09-22 18:46:57 +02:00

43
2

bfba6cfca8 · test(skill-eval): declare the Cursor caveat as a fixed token and grade it exactly · Updated 2026-09-22 09:16:02 +02:00

44
2

7807604239 · test(skill-eval): pin peer naming by requested model as a ce-pov cell · Updated 2026-09-22 09:10:11 +02:00

46
3

dbac59752b · fix(ce-plan): keep planning after the behavior trace returns · Updated 2026-09-22 08:57:58 +02:00

46
3

729cf2e445 · fix(ce-work): document omitted Grok effort as xhigh · Updated 2026-09-21 22:10:56 +02:00

50
2