Skip to content

v1.19.0

Choose a tag to compare

@agents-workflows-bot agents-workflows-bot released this 04 Aug 02:43
628a725

1.19.0 (2026-08-03)

Features

  • add typed capability effect evidence to agent runners (#2755) (b6c2324)
  • codex: promote Terra runner defaults (#2904) (cbd8673)
  • compile consumer sync into one typed normalized plan (#2756) (d1b3994)
  • deps: preserve agent repairs through promoted dependency PRs (#2853) (090a222)
  • llm: add evidence-gated model evaluation pipeline (#2710) (7a9e289)
  • llm: preflight model pilot candidates (885ef4e)
  • maint: batch canonical dev-tool updates (#2902) (4445fd7)
  • model-eval: per-provider preflight tolerance so one bad candidate can't abort the pilot (#2824) (9e04b91)
  • model-policy: exempt un-harvestable categories from the 10-case floor (#2873) (a84a558)
  • model-promotion: teach model_family a github-models publisher rule (#2855) (4464b6b)
  • pdf-extract: scaffold shared extraction package (46b1b3d)
  • publish bounded consumer-sync shadow evidence (#2757) (5aa070b)
  • renovate: bound fleet maintenance intake (#2893) (18f25a9)
  • renovate: stop consumer Renovate from editing Maint 68-owned files (#2892) (22f2393)
  • sync: gate fleet fan-out on consumer canaries (#2886) (40dd4e8)
  • sync: lease generated delivery attempts (#2895) (d1e5020)
  • sync: publish maintenance efficiency SLOs (#2896) (9c49bc6)
  • sync: reconcile generated delivery handoffs (#2898) (f425c5c)

Bug Fixes

  • address shared sync review blockers (#2778) (6e6c929)
  • align consumer model registry contract (#2775) (a7afe69)
  • align dev version sync PR body (#2733) (001bfed)
  • apply tracker:durable label when creating/updating weekly metrics tracker issue (8ae71a9)
  • baseline issue optimizer template drift (#2732) (8b528fc)
  • branch-protection: assert the reviewed policy instead of an aspirational one (#2888) (2321dbe)
  • branch-protection: stop requiring a conditionally-posted status context (#2872) (3bf9a1a)
  • clarify slot validation and config diagnostics (#2804) (01d9d86)
  • close model registry review gaps (#2780) (b1c7bcc)
  • close the review findings that landed after their PRs merged (#2890) (6b845c2)
  • consumers: vendor config into consumer LLM workflows (a87536c)
  • deps: route promotion PR source context (#2857) (8497346)
  • deps: update non-major updates (#2845) (097dd9d)
  • fail closed on empty LLM config overrides (#2800) (405eeaa)
  • harden auto-pilot stall handoffs (#2826) (31339e1)
  • harden consumer model slot resolution (#2785) (37477cb)
  • harden dependency repair promotion trust (#2867) (085c707)
  • harden synced model registry slots (#2794) (57a384d)
  • harden synced registry and keepalive helpers (#2793) (f2d175d)
  • harden synced registry contracts (#2806) (d492af9)
  • ignore tier slots with malformed registry (#2724) (057152b)
  • include agent registry in maint 68 checkout (#2774) (c8a93d2)
  • include manifest sources in sync sparse checkout (#2773) (34f273a)
  • keep template model registry fresh (#2719) (fd8335c)
  • llm-config: propagate the advanced selections to the consumer template (#2874) (3d57222)
  • llm: install provider clients for model pilot (5e14605)
  • llm: namespace github-models model ids for the GA endpoint (bare id → 404) (#2825) (fcf8c21)
  • llm: omit temperature for Anthropic models that reject it (unblocks verifier eval of Opus 4.8 / Claude 5) (#2822) (5275c48)
  • llm: run model evaluation pilot as a module (449f42d)
  • maint: close #2882 verifier CONCERNS gaps (#2903) (1e303fc)
  • meta: suppress issue links for dependency promotions (#2862) (f2d406b)
  • pin consumer LLM workflow installs (#2730) (87a5587)
  • port auto-label campaign guards to consumers (#2851) (9959061)
  • preserve consumer Ruff selections with 0.16 (#2840) (fdd8007)
  • preserve consumer slot compatibility (#2781) (7452d5a)
  • preserve CRLF for Windows command files (#2859) (1ad4624)
  • preserve model registry checker package import (#2809) (50e71db)
  • preserve reviewed model fallback (#2779) (a141a24)
  • preserve slot allowlist fail-closed behavior (#2797) (aff17b0)
  • refresh existing dev-tool sync branches (#2830) (5d25a44)
  • repair shared sync gate failures (#2722) (ff79b33)
  • request Actions capability for auto-pilot create-pr dispatches (#2870) (b232bb1)
  • request Actions capability for auto-pilot handoff (#2869) (81e5821)
  • restore backplane reference artifact (6391edf)
  • restore legacy registry sync compatibility (90948f1)
  • restore LLM slot bootstrap compatibility (#2796) (53f18ad)
  • retain ignored llm registry tier parameter (#2837) (885905e)
  • retain upgraded LLM provider pins (#2838) (cb78bf2)
  • scripts: drop dead run_tests.sh recommendation from check_test_dependencies.sh (#2835) (d362a6c)
  • sync all dev dependency pin surfaces (#2790) (4dd7de2)
  • sync: classify covered consumer drift (#2885) (79db012)
  • sync: close #2881 verifier CONCERNS gaps (#2900) (df80ffd)
  • sync: persist Maint 71 delivery handoffs (#2899) (5b7a926)
  • sync: safely refresh matching leased deliveries (#2901) (376c3d7)
  • tolerate missing durable tracker labels (#2772) (ffca68a)
  • unblock generated maintenance waves (#2789) (47304ed)
  • use non-cone checkout for runner base action (#2745) (f3036c1)
  • verifier: normalize structured provider content (#2887) (2e3e5b3)
  • verifier: vendor the LLM model registry into .workflows-lib (#2850) (fa2253e)

What's Changed

  • fix: keep template model registry fresh by @stranske in #2719
  • chore(deps): update stranske/workflows digest to ebef44a by @renovate[bot] in #2720
  • fix: repair shared sync gate failures by @stranske in #2722
  • feat(pdf-extract): scaffold single-source stranske-pdf-extract library (#2711) by @stranske in #2716
  • fix: ignore tier slots with malformed registry by @stranske in #2724
  • ci(pdf-extract): add package verification gate by @stranske in #2723
  • chore(deps): update dependency numpy to v2.5.1 by @renovate[bot] in #2727
  • chore(deps): update anthropics/claude-code-action digest to f87768c by @renovate[bot] in #2725
  • chore(deps): refresh pip-compile outputs by @renovate[bot] in #2728
  • fix: pin consumer LLM workflow installs by @stranske in #2730
  • chore: update shared coverage pin by @stranske in #2731
  • fix: baseline issue optimizer template drift by @stranske in #2732
  • fix: align dev version sync PR body by @stranske in #2733
  • chore(deps): update dependency langchain-core to v1.4.9 by @renovate[bot] in #2735
  • chore(deps): update non-major updates by @renovate[bot] in #2736
  • docs: quarterly README refresh sweep (2026-Q3) by @stranske in #2726
  • docs: update durable tracker pointer from #1796 to #2211 in ops docs by @stranske in #2737
  • fix: apply tracker:durable label when creating/updating weekly metrics tracker issue by @stranske in #2738
  • chore(coderabbit): focus reviews on core content — close Workflows blind spot + release filter + core-path guidance by @stranske in #2739
  • Fix agent runner base action sparse checkout by @stranske in #2745
  • Issue #2742: Propagate keepalive execution profiles by @stranske in #2746
  • Issue #2741: Add keepalive capability bundle contract by @stranske in #2744
  • Issue #2743: Record backplane registry evidence by @stranske in #2747
  • Add explicit Sol Terra Luna execution profiles by @stranske in #2748
  • Add immutable read-only Sol Terra Luna trial runner by @stranske in #2749
  • Update non-major updates by @renovate[bot] in #2754
  • feat: add typed capability effect evidence to agent runners by @stranske in #2755
  • feat: compile consumer sync into one typed normalized plan by @stranske in #2756
  • feat: publish bounded consumer-sync shadow evidence by @stranske in #2757
  • fix: restore backplane reference artifact by @stranske in #2763
  • feat(llm): add evidence-gated model evaluation pipeline (#2710) by @stranske in #2740
  • fix(llm): run model evaluation pilot as a module by @stranske in #2766
  • fix(llm): install provider clients for model pilot by @stranske in #2767
  • feat(llm): preflight model pilot candidates by @stranske in #2769
  • Update actions/setup-node action to v7 by @renovate[bot] in #2771
  • fix: tolerate missing durable tracker labels by @stranske in #2772
  • Refresh pip-compile outputs by @renovate[bot] in #2764
  • Update anthropics/claude-code-action digest to f1bd27c by @renovate[bot] in #2734
  • fix: include manifest sources in sync sparse checkout by @stranske in #2773
  • fix: include agent registry in maint 68 checkout by @stranske in #2774
  • fix: align consumer model registry contract by @stranske in #2775
  • chore(deps): update anthropics/claude-code-action digest to 1253134 by @renovate[bot] in #2777
  • chore(deps): sync coverage 7.15.1 pins by @stranske in #2776
  • fix: address shared sync review blockers by @stranske in #2778
  • fix: preserve reviewed model fallback by @stranske in #2779
  • fix: close model registry review gaps by @stranske in #2780
  • fix: preserve consumer slot compatibility by @stranske in #2781
  • chore(deps): update anthropics/claude-code-action digest to 1298632 by @renovate[bot] in #2783
  • fix: harden consumer model slot resolution by @stranske in #2785
  • chore(deps): update dependency langchain to v1.3.14 by @renovate[bot] in #2786
  • chore(deps): update anthropics/claude-code-action digest to 700e7f8 by @renovate[bot] in #2787
  • chore(deps): update dependency tomlkit to v0.15.1 by @renovate[bot] in #2788
  • chore: update dev tool versions from PyPI by @github-actions[bot] in #2784
  • fix: unblock generated maintenance waves by @stranske in #2789
  • chore(deps): update anthropics/claude-code-action digest to 3553f84 by @renovate[bot] in #2791
  • fix: sync all dev dependency pin surfaces by @stranske in #2790
  • chore(deps): update astral-sh/setup-uv action to v8 by @renovate[bot] in #2765
  • chore(deps): update anthropics/claude-code-action digest to af0559e by @renovate[bot] in #2792
  • fix: harden synced registry and keepalive helpers by @stranske in #2793
  • fix: harden synced model registry slots by @stranske in #2794
  • fix: restore LLM slot bootstrap compatibility by @stranske in #2796
  • fix: preserve slot allowlist fail-closed behavior by @stranske in #2797
  • docs: clarify LLM slot bootstrap condition by @stranske in #2798
  • chore(deps): update actions/checkout action to v7.0.1 by @renovate[bot] in #2799
  • fix: fail closed on empty LLM config overrides by @stranske in #2800
  • chore(deps): update astral-sh/setup-uv action to v9 by @renovate[bot] in #2803
  • chore(deps): update anthropics/claude-code-action digest to fa7e2f0 by @renovate[bot] in #2801
  • fix: clarify slot validation and config diagnostics by @stranske in #2804
  • fix: harden synced registry contracts by @stranske in #2806
  • fix: restore legacy registry sync compatibility by @stranske in #2807
  • fix: preserve model registry checker package import by @stranske in #2809
  • chore(deps): update ossf/scorecard-action action to v2.4.4 by @renovate[bot] in #2811
  • Bounded agent stall rotation before needs-human (auto-pilot) by @stranske in #2814
  • Re-baseline Health 74 template-drift fingerprints (fix red main) by @stranske in #2815
  • Stop backplane reference-run staleness from blocking all PRs (Health/summary) by @stranske in #2821
  • Make Health 74 template-drift check pin-tolerant (stop the recurrence) by @stranske in #2817
  • fix(llm): omit temperature for Anthropic models that reject it (unblocks verifier eval of Opus 4.8 / Claude 5) by @stranske in #2822
  • feat(model-eval): per-provider preflight tolerance (one bad candidate no longer aborts the pilot) — part of #2819 by @stranske in #2824
  • chore(model-eval): refresh verifier pilot candidates to current catalog (add claude-opus-4-8) by @stranske in #2818
  • Backplane reference-run freshness: warning-only + weekly cadence (self-heal CI) by @stranske in #2827
  • fix(llm): namespace github-models model ids for the GA endpoint (bare id → 404) by @stranske in #2825
  • chore(deps): update anthropics/claude-code-action digest to e0cf66d by @renovate[bot] in #2808
  • chore(deps): update anthropics/claude-code-action digest to be7b93b by @renovate[bot] in #2829
  • freshness gate: overdue review is advisory, not a fleet-wide hard block by @stranske in #2833
  • maint-78: derive pilot candidates from the registry + auto-trigger (#2819 move 1) by @stranske in #2831
  • maint-79: grow verifier corpus from realized PR outcomes (#2819 move 2) by @stranske in #2832
  • maint-86: prepare (not auto-apply) verifier model promotions (#2819 move 3) by @stranske in #2834
  • fix: refresh existing dev-tool sync branches by @stranske in #2830
  • fix(scripts): drop dead run_tests.sh recommendation from check_test_dependencies.sh by @stranske in #2835
  • fix: retain ignored llm registry tier parameter by @stranske in #2837
  • fix: retain upgraded LLM provider pins by @stranske in #2838
  • chore(deps): refresh pip-compile outputs by @renovate[bot] in #2839
  • fix: preserve consumer Ruff selections with 0.16 by @stranske in #2840
  • corpus: harvest realized-outcome verifier cases by @stranske in #2843
  • fix(deps): update non-major updates by @renovate[bot] in #2845
  • chore(deps): update non-major updates by @renovate[bot] in #2846
  • chore(deps): update dependency langchain-core to v1.5.3 by @renovate[bot] in #2847
  • fix(verifier): vendor the LLM model registry into .workflows-lib by @stranske in #2850
  • chore: update dev tool versions from PyPI by @github-actions[bot] in #2849
  • fix: port auto-label campaign guards to consumers by @stranske in #2851
  • feat(deps): preserve agent repairs through promoted dependency PRs by @stranske in #2853
  • chore: untrack the per-run langsmith worker-attempt artifact by @stranske in #2856
  • feat(model-promotion): teach model_family a github-models publisher rule by @stranske in #2855
  • chore(model-registry): advance provisional verifier selections to current catalog by @stranske in #2852
  • fix(deps): route promotion PR source context by @stranske in #2857
  • fix: preserve CRLF for Windows command files by @stranske in #2859
  • test(actions): make setup-python contract upgrade-safe by @stranske in #2854
  • test(actions): close setup-python v7 review gaps by @stranske in #2860
  • chore(deps): update actions/setup-python action to v7 by @renovate[bot] in #2795
  • fix(consumers): vendor config into the consumer LLM-client workflows by @stranske in #2861
  • fix(meta): suppress issue links for dependency promotions by @stranske in #2862
  • test(source-context): cover partial-label and mismatched promotion markers by @stranske in #2866
  • fix: harden dependency repair promotion trust by @stranske in #2867
  • fix: harden auto-pilot stall handoffs by @stranske in #2826
  • fix: request Actions capability for auto-pilot handoff by @stranske in #2869
  • fix: request Actions capability for auto-pilot create-pr dispatches by @stranske in #2870
  • docs(ops): retarget durable-tracker table at the live tracker issues by @stranske in #2871
  • fix(branch-protection): stop requiring a conditionally-posted status context by @stranske in #2872
  • fix(llm-config): propagate the advanced selections to the consumer template by @stranske in #2874
  • feat(model-policy): exempt un-harvestable categories from the 10-case floor by @stranske in #2873
  • docs(contracts): stop synced contract docs linking to unsynced siblings by @stranske in #2875
  • fix(sync): classify covered consumer drift by @stranske in #2885
  • chore: remove the one-off testgen acceptance gate for closed issue #2620 by @stranske in #2889
  • fix(branch-protection): assert the reviewed policy instead of an aspirational one by @stranske in #2888
  • fix(verifier): normalize structured provider content by @stranske in #2887
  • feat(sync): gate fleet fan-out on consumer canaries by @stranske in #2886
  • docs(ops): give #2210 the drift-state contract #2878 required by @stranske in #2891
  • feat(renovate): bound fleet maintenance intake by @stranske in #2893
  • feat(sync): lease generated delivery attempts by @stranske in #2895
  • feat(sync): publish maintenance efficiency SLOs by @stranske in #2896
  • feat(sync): reconcile generated delivery handoffs by @stranske in #2898
  • fix(sync): persist Maint 71 delivery handoffs by @stranske in #2899
  • fix(sync): close #2881 verifier CONCERNS gaps by @stranske in #2900
  • fix(sync): safely refresh matching leased deliveries by @stranske in #2901
  • fix: close the review findings that landed after their PRs merged by @stranske in #2890
  • feat(renovate): stop consumer Renovate from editing Maint 68-owned files by @stranske in #2892
  • feat(maint): batch canonical dev-tool updates by @stranske in #2902
  • fix(maint): close #2882 verifier CONCERNS gaps by @stranske in #2903
  • feat(codex): promote Terra runner defaults by @stranske in #2904
  • corpus: harvest realized-outcome verifier cases by @stranske in #2906
  • fix(verifier): repair #2904 review residue by @stranske in #2907
  • chore(main): release 1.19.0 by @agents-workflows-bot[bot] in #2721

Full Changelog: v1.18.0...v1.19.0