@open-agent-toolkit/cli 0.3.0 → 0.3.2

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (76) hide show
  1. package/assets/agents/oat-reviewer.md +10 -3
  2. package/assets/bundle-metadata.json +1 -1
  3. package/assets/config/dispatch-matrix-recommendation.json +63 -43
  4. package/assets/docs/cli-utilities/configuration.md +20 -8
  5. package/assets/docs/cli-utilities/workflow-gates.md +1 -1
  6. package/assets/docs/contributing/index.md +1 -0
  7. package/assets/docs/contributing/updating-model-guidance.md +106 -0
  8. package/assets/docs/contributing/verifying-cursor-pins.md +30 -18
  9. package/assets/docs/provider-sync/providers.md +5 -2
  10. package/assets/docs/reference/cli-reference.md +1 -1
  11. package/assets/docs/reference/oat-directory-structure.md +1 -1
  12. package/assets/docs/workflows/projects/artifacts.md +1 -1
  13. package/assets/docs/workflows/projects/dispatch-ceiling.md +67 -50
  14. package/assets/docs/workflows/projects/evidence-layers.md +6 -2
  15. package/assets/docs/workflows/projects/implementation-execution.md +9 -4
  16. package/assets/docs/workflows/projects/lifecycle.md +8 -6
  17. package/assets/docs/workflows/projects/programmatic-execution.md +4 -3
  18. package/assets/public-package-versions.json +4 -4
  19. package/assets/skills/oat-dispatch-subagents/SKILL.md +1 -1
  20. package/assets/skills/oat-dispatch-subagents/references/provider-claude.md +20 -7
  21. package/assets/skills/oat-project-autonomous/references/gate-inventory.md +4 -4
  22. package/assets/skills/oat-project-document/references/docs/autonomy-contract.md +4 -4
  23. package/assets/skills/oat-project-implement/SKILL.md +1 -1
  24. package/assets/skills/oat-project-implement/references/completion-and-closeout.md +2 -0
  25. package/assets/skills/oat-project-implement/references/dispatch-and-dry-run.md +82 -25
  26. package/assets/skills/oat-project-implement/references/docs/autonomy-contract.md +4 -4
  27. package/assets/skills/oat-project-implement/references/phase-execution.md +16 -8
  28. package/assets/skills/oat-project-import-plan/SKILL.md +7 -4
  29. package/assets/skills/oat-project-lite/SKILL.md +5 -3
  30. package/assets/skills/oat-project-lite/references/docs/autonomy-contract.md +4 -4
  31. package/assets/skills/oat-project-plan/SKILL.md +7 -4
  32. package/assets/skills/oat-project-plan-writing/SKILL.md +28 -16
  33. package/assets/skills/oat-project-pr-final/references/docs/autonomy-contract.md +4 -4
  34. package/assets/skills/oat-project-quick-start/SKILL.md +7 -4
  35. package/assets/skills/oat-project-quick-start/references/docs/autonomy-contract.md +4 -4
  36. package/assets/skills/oat-project-review-provide/SKILL.md +7 -4
  37. package/assets/skills/subagent-orchestration/SKILL.md +1 -1
  38. package/assets/skills/subagent-orchestration/references/evidence-and-refresh.md +30 -29
  39. package/assets/skills/subagent-orchestration/references/provider-claude.md +41 -36
  40. package/assets/skills/subagent-orchestration/references/provider-codex.md +30 -35
  41. package/assets/skills/subagent-orchestration/references/provider-cursor.md +46 -38
  42. package/assets/templates/plan.md +5 -5
  43. package/dist/commands/gate/review-verdict.d.ts.map +1 -1
  44. package/dist/commands/gate/review-verdict.js +23 -1
  45. package/dist/commands/project/dispatch/record.d.ts.map +1 -1
  46. package/dist/commands/project/dispatch/record.js +27 -5
  47. package/dist/commands/project/dispatch-ceiling/index.d.ts.map +1 -1
  48. package/dist/commands/project/dispatch-ceiling/index.js +112 -35
  49. package/dist/config/dispatch-matrix.d.ts.map +1 -1
  50. package/dist/config/dispatch-matrix.js +13 -0
  51. package/dist/providers/ceiling/registry.d.ts +7 -2
  52. package/dist/providers/ceiling/registry.d.ts.map +1 -1
  53. package/dist/providers/ceiling/registry.js +35 -11
  54. package/dist/providers/claude/codec/materialize.d.ts +33 -0
  55. package/dist/providers/claude/codec/materialize.d.ts.map +1 -0
  56. package/dist/providers/claude/codec/materialize.js +149 -0
  57. package/dist/providers/claude/codec/sync-extension.d.ts +24 -0
  58. package/dist/providers/claude/codec/sync-extension.d.ts.map +1 -0
  59. package/dist/providers/claude/codec/sync-extension.js +312 -0
  60. package/dist/providers/claude/dispatch-envelope.d.ts +34 -0
  61. package/dist/providers/claude/dispatch-envelope.d.ts.map +1 -0
  62. package/dist/providers/claude/dispatch-envelope.js +233 -0
  63. package/dist/providers/claude/targets.d.ts +48 -0
  64. package/dist/providers/claude/targets.d.ts.map +1 -0
  65. package/dist/providers/claude/targets.js +228 -0
  66. package/dist/providers/codex/codec/shared.d.ts +6 -0
  67. package/dist/providers/codex/codec/shared.d.ts.map +1 -1
  68. package/dist/providers/codex/codec/shared.js +8 -0
  69. package/dist/providers/cursor/codec/catalog.d.ts.map +1 -1
  70. package/dist/providers/cursor/codec/catalog.js +47 -45
  71. package/dist/providers/identity/claude-runtime-observation.fixtures.d.ts +35 -0
  72. package/dist/providers/identity/claude-runtime-observation.fixtures.d.ts.map +1 -1
  73. package/dist/providers/identity/claude-runtime-observation.fixtures.js +225 -12
  74. package/dist/providers/shared/registry.d.ts.map +1 -1
  75. package/dist/providers/shared/registry.js +5 -2
  76. package/package.json +2 -2
package/assets/skills/oat-project-quick-start/references/docs/autonomy-contract.md CHANGED
@@ -253,7 +253,7 @@ remains immutable baseline evidence for the original p01-t01 verification.
253
253
  | `oat-project-import-plan/SKILL.md` | `f7ca8635452f -> NG`; `c2e5fa410069 -> NG`; `655d61df0a8f -> NG`; `dceff5cc04b5 -> NG`; `1af4957a1af2 -> IMPORT-01`; `45b699c1c0b5 -> IMPORT-01`; `ba6d120791c1 -> IMPORT-01`; `070ebc5259af -> IMPORT-03`; `c11552696e2e -> IMPORT-03`; `cfe661966baf -> IMPORT-04`; `baa91c0a0bf8 -> NG`; `e364e84b2702 -> NG`; `39031985a945 -> IMPORT-06`; `f106aa29a217 -> IMPORT-07`; `65038450aee3 -> IMPORT-07`; `bee4c4aa34c6 -> NG`; `3f0b53056edd -> IMPORT-09`; `30aa74677c8d -> IMPORT-03`; `430c2bc030a6 -> IMPORT-03`; `93522f67da52 -> NG` |
254
254
  | `oat-project-implement/SKILL.md` | `18980f6bc1a6 -> NG`; `e0355c7a05e9 -> IMPLEMENT-11+IMPLEMENT-13+IMPLEMENT-16`; `6296f41ae11c -> NG`; `a167d03559fa -> NG`; `cf317c17c586 -> IMPLEMENT-01`; `8ce8a09feca4 -> NG`; `bd0cb4988188 -> NG`; `28af4ee0055d -> NG` |
255
255
  | `oat-project-implement/references/completion-and-closeout.md` | `87f1424a8692 -> NG`; `6b1ec05182f0 -> IMPLEMENT-11`; `e6d696a27c91 -> NG`; `c77f8ccb52ad -> NG`; `48de2614c0ba -> IMPLEMENT-11`; `32d53f5212e5 -> IMPLEMENT-11`; `7cdb4a9c57b7 -> IMPLEMENT-11`; `16ffc0aa099b -> IMPLEMENT-11`; `d6868ea10b3f -> IMPLEMENT-11`; `4f05ff5567be -> IMPLEMENT-11`; `c70b47ae7a6d -> IMPLEMENT-11`; `5fafd093b58e -> NG`; `21ceaf7e658a -> NG`; `249427a9f475 -> NG`; `8a7ec13a9b7c -> NG`; `9aeb7fce2012 -> IMPLEMENT-18`; `f11a7cdda220 -> IMPLEMENT-18`; `4523ac029152 -> NG`; `6e87b08c4046 -> NG`; `baa6052c7f0a -> IMPLEMENT-18`; `27d3fad116b1 -> IMPLEMENT-18`; `2cfbeb4388a5 -> IMPLEMENT-18`; `c56ea56a56db -> NG`; `43e2e3c1380d -> IMPLEMENT-17`; `b49690277fb5 -> NG`; `df12304d6a36 -> NG`; `52eacac11fef -> IMPLEMENT-17`; `60092df6b17b -> IMPLEMENT-17`; `361872c6b00e -> IMPLEMENT-17`; `d2bc711a75fb -> IMPLEMENT-17`; `9889538f6212 -> NG`; `58add7778cef -> NG`; `3984cd158ca8 -> NG`; `64a638393159 -> NG`; `ecdaaf6c76fa -> NG`; `96e01b7c0713 -> NG`; `79806948373a -> NG`; `82701dfaf962 -> NG`; `b1c7b8299bf7 -> IMPLEMENT-13+IMPLEMENT-15`; `352caf79e5cb -> IMPLEMENT-13+IMPLEMENT-15`; `6792dd0068a7 -> IMPLEMENT-13+IMPLEMENT-15`; `91f4ebf1704c -> IMPLEMENT-13+IMPLEMENT-15`; `1119540c90cd -> IMPLEMENT-13+IMPLEMENT-15`; `13e6d608682c -> IMPLEMENT-15`; `9fc3aa86e096 -> IMPLEMENT-13+IMPLEMENT-15`; `b44bd8f4fe7e -> NG`; `d62f3ae68677 -> NG`; `4133bca9bd9f -> IMPLEMENT-12`; `f6f87d97896b -> IMPLEMENT-14`; `9d066a59d00a -> IMPLEMENT-14`; `abd7a8d8fc89 -> IMPLEMENT-14`; `a2199f9dafee -> IMPLEMENT-14`; `fcef83b91326 -> NG`; `e2df750966f5 -> IMPLEMENT-13+IMPLEMENT-15`; `2b714060ccae -> IMPLEMENT-13+IMPLEMENT-15`; `708e8e6d33ac -> IMPLEMENT-15`; `f81e67549c28 -> IMPLEMENT-13+IMPLEMENT-15`; `8980233e250d -> NG`; `713729395d08 -> IMPLEMENT-13`; `671f3622cd05 -> IMPLEMENT-13`; `19c38f351f3e -> IMPLEMENT-13+IMPLEMENT-14+IMPLEMENT-15+IMPLEMENT-16`; `b552eeb1fa37 -> IMPLEMENT-13+IMPLEMENT-14+IMPLEMENT-15+IMPLEMENT-16`; `24de52a2e140 -> IMPLEMENT-13+IMPLEMENT-14+IMPLEMENT-15+IMPLEMENT-16`; `18071a1287a9 -> IMPLEMENT-13+IMPLEMENT-14+IMPLEMENT-15+IMPLEMENT-16`; `318989f76cd7 -> IMPLEMENT-16`; `50c9170a3fc8 -> NG`; `8af4d1f8c32c -> IMPLEMENT-13+IMPLEMENT-14+IMPLEMENT-15+IMPLEMENT-16`; `f47aefb3a3cc -> IMPLEMENT-16`; `7c7bcd0381cb -> NG`; `4808cd2e14d6 -> NG`; `906d7d8628af -> IMPLEMENT-13+IMPLEMENT-14+IMPLEMENT-15+IMPLEMENT-16`; `c7221ba40303 -> IMPLEMENT-13+IMPLEMENT-14+IMPLEMENT-15+IMPLEMENT-16`; `05eaa2d6f58f -> IMPLEMENT-13+IMPLEMENT-14+IMPLEMENT-15+IMPLEMENT-16`; `acf05140d8f3 -> IMPLEMENT-19`; `6d68053180a0 -> NG`; `1128ae7838f5 -> IMPLEMENT-19`; `2b3e427b5fc0 -> IMPLEMENT-19`; `9fdeb38897a2 -> IMPLEMENT-13+IMPLEMENT-15`; `4f1c82e4bef8 -> IMPLEMENT-13+IMPLEMENT-15`; `d74b5b3c4ad3 -> IMPLEMENT-13+IMPLEMENT-14+IMPLEMENT-15+IMPLEMENT-16`; `85b8fae4aa24 -> IMPLEMENT-13+IMPLEMENT-14+IMPLEMENT-15+IMPLEMENT-16`; `912923a29ecf -> IMPLEMENT-13+IMPLEMENT-14+IMPLEMENT-15+IMPLEMENT-16`; `c9707e20ad4a -> IMPLEMENT-13+IMPLEMENT-14+IMPLEMENT-15+IMPLEMENT-16`; `8108196d68a5 -> IMPLEMENT-13+IMPLEMENT-14+IMPLEMENT-15+IMPLEMENT-16`; `ad9d8e0114a6 -> IMPLEMENT-16`; `e6ef8cdc8986 -> IMPLEMENT-16`; `d3891824ecfa -> IMPLEMENT-13+IMPLEMENT-14+IMPLEMENT-15+IMPLEMENT-16`; `045d486db046 -> IMPLEMENT-13+IMPLEMENT-14+IMPLEMENT-15+IMPLEMENT-16`; `7f3891ba6f9d -> IMPLEMENT-13+IMPLEMENT-14+IMPLEMENT-15+IMPLEMENT-16`; `2db46931ac00 -> IMPLEMENT-13+IMPLEMENT-14+IMPLEMENT-15+IMPLEMENT-16`; `aaaf9a6d018a -> IMPLEMENT-16`; `5198341f296c -> IMPLEMENT-16`; `4443b0d16d43 -> IMPLEMENT-16`; `76511502e95b -> IMPLEMENT-16`; `585a1f4575a1 -> IMPLEMENT-16`; `7dc4e4afca1e -> IMPLEMENT-16`; `0d3f1d6038fd -> IMPLEMENT-16`; `e3dfc24fbed6 -> IMPLEMENT-16`; `f19c8013230c -> IMPLEMENT-16`; `ba83e7ce6aed -> IMPLEMENT-16`; `4e4c5485b755 -> IMPLEMENT-16`; `cc8e7e01f30c -> IMPLEMENT-17`; `a13b8adde5da -> IMPLEMENT-17` |
256
- | `oat-project-implement/references/dispatch-and-dry-run.md` | `21b85f5e4daa -> IMPLEMENT-08`; `74c92b32bb2a -> IMPLEMENT-08`; `8efd2d0517ab -> IMPLEMENT-09`; `461ed9b3e495 -> IMPLEMENT-09`; `b5ad64fd744d -> NG`; `14ce436cfffa -> NG`; `043ca36da3e2 -> NG`; `a23e45c1dc8d -> NG`; `498143bb41a7 -> NG`; `4fe3eb179023 -> NG`; `7f75485e5a33 -> NG`; `8ec3a96df029 -> NG`; `16c3f96e758d -> NG`; `5915c41fe058 -> NG`; `f51e43eeac56 -> NG`; `fd41c25623cb -> NG`; `58a932ed6453 -> NG`; `4718bd3ce12d -> NG`; `58c444a2126b -> NG`; `2df3adae4120 -> NG`; `e75df892b55c -> NG`; `8ba5dbd2a0d8 -> NG`; `4d6c519131e5 -> NG` |
256
+ | `oat-project-implement/references/dispatch-and-dry-run.md` | `21b85f5e4daa -> IMPLEMENT-08`; `74c92b32bb2a -> IMPLEMENT-08`; `8efd2d0517ab -> IMPLEMENT-09`; `461ed9b3e495 -> IMPLEMENT-09`; `b5ad64fd744d -> NG`; `14ce436cfffa -> NG`; `043ca36da3e2 -> NG`; `a23e45c1dc8d -> NG`; `498143bb41a7 -> NG`; `4fe3eb179023 -> NG`; `7f75485e5a33 -> NG`; `8ec3a96df029 -> NG`; `16c3f96e758d -> NG`; `5915c41fe058 -> NG`; `f51e43eeac56 -> NG`; `fd41c25623cb -> NG`; `58a932ed6453 -> NG`; `4718bd3ce12d -> NG`; `58c444a2126b -> NG`; `e75df892b55c -> NG`; `8ba5dbd2a0d8 -> NG`; `4d6c519131e5 -> NG` |
257
257
  | `oat-project-implement/references/docs/autonomy-contract.md` | `* -> NG` |
258
258
  | `oat-project-implement/references/phase-execution.md` | `27737e814790 -> NG`; `6dda2b8e62e2 -> IMPLEMENT-10`; `b3197086bb1e -> IMPLEMENT-10`; `25f2a6e104ee -> NG`; `4aa21120295f -> NG` |
259
259
  | `oat-project-implement/references/plan-and-resume.md` | `413829d31e32 -> IMPLEMENT-02`; `c7706cd8ad09 -> IMPLEMENT-03`; `b7ea139780ec -> IMPLEMENT-03+IMPLEMENT-04`; `8d438f2180ba -> IMPLEMENT-03`; `9e72e0518cda -> IMPLEMENT-03`; `3e177a1021e6 -> IMPLEMENT-03+IMPLEMENT-04`; `b59fc9eb9ce3 -> IMPLEMENT-03`; `a2b1d6855283 -> IMPLEMENT-03`; `44351a32e52a -> IMPLEMENT-03`; `b98b022f5523 -> IMPLEMENT-03`; `e9f5b949e214 -> IMPLEMENT-03`; `316fb6d9e980 -> IMPLEMENT-03`; `8b25a2c5c548 -> IMPLEMENT-03`; `1a59b1c0e705 -> IMPLEMENT-03`; `466cd562fc52 -> IMPLEMENT-03`; `083d896978d5 -> IMPLEMENT-03`; `f4ab11c13ae9 -> IMPLEMENT-03`; `7dc17ffcc1c4 -> IMPLEMENT-03`; `f731a90adabd -> IMPLEMENT-04`; `886ac01e7aa9 -> IMPLEMENT-04`; `b715db67df0c -> IMPLEMENT-04`; `2d9f26a539f3 -> IMPLEMENT-10`; `28384dee114f -> IMPLEMENT-05`; `f2591a4978b4 -> NG`; `8290823080b5 -> IMPLEMENT-06`; `f75789bf6c8e -> IMPLEMENT-07`; `b5bd4852c58d -> NG`; `6e7fac10d436 -> NG`; `16f6203aadac -> NG`; `4b876e81abad -> NG` |
@@ -280,9 +280,9 @@ remains immutable baseline evidence for the original p01-t01 verification.
280
280
  | `oat-project-complete/SKILL.md` | `0177956707aa -> NG`; `8305b5bb65ef -> NG`; `3c0b26f0ff1c -> NG`; `8cc1fa5aab1e -> COMPLETE-01`; `4c209fbb4716 -> COMPLETE-01`; `02e4175e1171 -> COMPLETE-01`; `d698d4f98e76 -> COMPLETE-01`; `9f86a7218d93 -> COMPLETE-01`; `024713698d9a -> COMPLETE-01`; `c327995053fa -> COMPLETE-01`; `61f81d4ec7c1 -> COMPLETE-01`; `14fb2e56809f -> COMPLETE-01`; `7d4ff668744d -> COMPLETE-01`; `502a6c141784 -> COMPLETE-01`; `fb7e2db356af -> COMPLETE-01`; `f73f8476360a -> NG`; `1e15f420039f -> NG`; `cc234032ddd7 -> NG`; `1dd54687229b -> NG`; `fc05c8303cd2 -> NG`; `24be721d635e -> NG` |
281
281
  | `oat-project-lite/references/docs/autonomy-contract.md` | `* -> NG` |
282
282
  | `oat-project-implement/references/completion-and-closeout.md` | `87f1424a8692 -> NG`; `6b1ec05182f0 -> IMPLEMENT-11`; `e6d696a27c91 -> NG`; `c77f8ccb52ad -> NG`; `48de2614c0ba -> IMPLEMENT-11`; `32d53f5212e5 -> IMPLEMENT-11`; `7cdb4a9c57b7 -> IMPLEMENT-11`; `16ffc0aa099b -> IMPLEMENT-11`; `d6868ea10b3f -> IMPLEMENT-11`; `4f05ff5567be -> IMPLEMENT-11`; `c70b47ae7a6d -> IMPLEMENT-11`; `5fafd093b58e -> NG`; `21ceaf7e658a -> NG`; `249427a9f475 -> NG`; `8a7ec13a9b7c -> NG`; `9aeb7fce2012 -> IMPLEMENT-18`; `f11a7cdda220 -> IMPLEMENT-18`; `4523ac029152 -> NG`; `6e87b08c4046 -> NG`; `baa6052c7f0a -> IMPLEMENT-18`; `27d3fad116b1 -> IMPLEMENT-18`; `2cfbeb4388a5 -> IMPLEMENT-18`; `c56ea56a56db -> NG`; `43e2e3c1380d -> IMPLEMENT-17`; `b49690277fb5 -> NG`; `df12304d6a36 -> NG`; `52eacac11fef -> IMPLEMENT-17`; `60092df6b17b -> IMPLEMENT-17`; `361872c6b00e -> IMPLEMENT-17`; `d2bc711a75fb -> IMPLEMENT-17` |
283
- | `oat-project-quick-start/SKILL.md` | `4f393afc3b4d -> QS-13`; `40ddcac880ef -> NG` |
284
- | `oat-project-plan/SKILL.md` | `2b32bc5e9fea -> NG`; `2f6c4fa8b6f9 -> PLAN-13`; `9fba3b40b522 -> NG` |
285
- | `oat-project-import-plan/SKILL.md` | `4bda28585b46 -> IMPORT-10`; `8ea53b60c725 -> NG` |
283
+ | `oat-project-quick-start/SKILL.md` | `aa560418765a -> QS-13`; `40ddcac880ef -> NG` |
284
+ | `oat-project-plan/SKILL.md` | `2b32bc5e9fea -> NG`; `58cc6b0adf97 -> PLAN-13`; `9fba3b40b522 -> NG` |
285
+ | `oat-project-import-plan/SKILL.md` | `349790ade2e1 -> IMPORT-10`; `8ea53b60c725 -> NG` |
286
286
  | `oat-project-implement/references/completion-and-closeout.md` | `5636ed3a0418 -> NG` |
287
287
 
288
288
  ## Prompt-scan comparison
package/assets/skills/oat-project-review-provide/SKILL.md CHANGED
@@ -5,7 +5,7 @@ disable-model-invocation: false
5
5
  user-invocable: true
6
6
  allowed-tools: Read, Glob, Grep, Write, Edit, Bash(git:*), Bash(oat:*), Bash(pnpm:*), Bash(mkdir:*), Bash(date:*), Bash(realpath:*), Bash(awk:*), AskUserQuestion
7
7
  metadata:
8
- version: 1.5.9
8
+ version: 1.5.10
9
9
  ---
10
10
 
11
11
  # Request Review
@@ -720,8 +720,11 @@ Build the actual provider invocation before reporting the target as enforced:
720
720
  `${WORKFLOWS_AGENT_PROVIDER_ROOT}/agents/oat-reviewer.md`,
721
721
  and the same Review Scope payload. If neither exact route is available, use
722
722
  only a verified-equivalent inline route or block the review.
723
- - Claude requires a non-empty `providers.claude.dispatchArgs.model`; the actual
724
- provider invocation must include that exact value as its `model` argument.
723
+ - Claude requires the exact `providers.claude.dispatchArgs.variant` for an
724
+ effort-pinned target; launch that generated native agent type and take effort
725
+ from its frontmatter. A legacy model-only target instead requires the exact
726
+ `providers.claude.dispatchArgs.model`. Any model supplied with a variant must
727
+ match its definition.
725
728
  - Cursor requires a non-empty `providers.cursor.dispatchArgs.variant`; the
726
729
  actual provider invocation must launch that exact resolver-returned native
727
730
  reviewer variant as the native agent type first. Keep Cursor model strings
@@ -875,7 +878,7 @@ First, pre-compute the review artifact path using Step 7 naming conventions so i
875
878
  Then spawn the reviewer:
876
879
 
877
880
  - Use provider-appropriate dispatch:
878
- - Claude Code: Task tool with `subagent_type: "oat-reviewer"` (resolves from `.claude/agents/oat-reviewer.md`). For a concrete managed target, the payload must also contain `model: providers.claude.dispatchArgs.model` with the resolver-returned value.
881
+ - Claude Code: for an effort-pinned target, use the exact generated `providers.claude.dispatchArgs.variant` as the native agent type. For a legacy model-only target, use `subagent_type: "oat-reviewer"` and include the exact resolver-returned `model: providers.claude.dispatchArgs.model`. The Agent call has no effort field.
879
882
  - Cursor: for a concrete managed target, invoke `providers.cursor.dispatchArgs.variant` as the exact resolver-selected native reviewer variant. Do not attach a Task-level model argument or normalize the mapped model. Base `oat-reviewer` is allowed only for explicit inherit/default behavior. A pre-start native role-selection rejection is the only replacement boundary.
880
883
  - Codex style: for a concrete managed target, first spawn the exact resolver-returned native `agent_type`; only an explicit pre-start native role-selection rejection permits the explicitly pinned fresh-child route from Step 6.0. Generic auto-selection is permitted only for the documented base-role exceptions.
881
884
  - Pass the Review Scope metadata block from Step 5 as the prompt
package/assets/skills/subagent-orchestration/SKILL.md CHANGED
@@ -4,7 +4,7 @@ description: Use when delegating work to subagents or choosing a model for a tas
4
4
  compatibility: Self-contained; no OAT installation required.
5
5
  user-invocable: true
6
6
  metadata:
7
- version: 1.0.4
7
+ version: 1.0.8
8
8
  ---
9
9
 
10
10
  # Subagent Orchestration
package/assets/skills/subagent-orchestration/references/evidence-and-refresh.md CHANGED
@@ -1,9 +1,9 @@
1
1
  # Evidence and Refresh Policy
2
2
 
3
3
  ```yaml
4
- guidance_version: 2026-07-25
5
- last_verified: 2026-07-25
6
- review_after: 2026-09-08
4
+ guidance_version: 2026-09-23
5
+ last_verified: 2026-09-23
6
+ review_after: 2026-10-08
7
7
  stale_after: 2026-10-23
8
8
  ```
9
9
 
@@ -17,11 +17,11 @@ authorization boundary.
17
17
  - `review-required`: older than 45 days, a newer family is observed, or a material control, price, tool, context, alias, or harness change is detected.
18
18
  - `stale`: older than 90 days, named models are unavailable or deprecated, or the launching surface can no longer express the recorded controls.
19
19
 
20
- Provider mappings in this package become review-required on 2026-09-08 and
21
- stale on 2026-10-23 unless re-verified earlier, or on the earlier date a
22
- reference states in its own header. `provider-cursor.md` carries an earlier
23
- 2026-09-04 review date because its catalog was not re-verified in the
24
- 2026-07-25 refresh.
20
+ The named routes were refreshed against provider catalogues on 2026-09-23.
21
+ The benchmark observations below remain dated to 2026-07-25 and require
22
+ review before making a new cross-provider performance claim. Each provider
23
+ reference carries its own verification date; availability refresh does not
24
+ turn earlier benchmark comparisons into fresh measurements.
25
25
 
26
26
  ## Immediate Review Triggers
27
27
 
@@ -78,15 +78,16 @@ publish a provider-independent effort conversion.
78
78
 
79
79
  ## Current Evidence Summary
80
80
 
81
- - OpenAI positions GPT-5.6 Sol as frontier, Terra as balanced, and Luna as
82
- high-volume. Official guidance starts at medium generally, but independent
83
- coding-agent data shows meaningful Luna/Terra quality cliffs below high for
84
- repository work. The dated mechanical Codex floor is therefore Luna high.
85
- - Post-release Opus 5 evidence supports medium for normal substantive Claude
86
- work and high for hard reasoning. Use xhigh for a reasoning-depth bottleneck
87
- or evaluated long-horizon gain; max is selective and requires an effort
88
- sweep. Consequence adds independent review rather than automatically raising
89
- effort.
81
+ - GPT-6 Sol and Luna are selectable in the current Codex catalogue and are
82
+ the default new-work routes in `provider-codex.md`. GPT-5.6 Sol/Luna/Terra
83
+ remain supported for explicit configurations. The July independent coding
84
+ data on GPT-5.6 Luna/Terra supported a mechanical floor of Luna high; this
85
+ is historical evidence, not a measured GPT-6 performance comparison.
86
+ - Opus 5.5 is the current substantive Claude route. Start at medium for
87
+ normal work and high for hard reasoning; xhigh and max require task-specific
88
+ evidence. Earlier Opus 5 effort observations are historical and do not
89
+ establish an Opus 5.5 benchmark result. Consequence adds independent review
90
+ rather than automatically raising effort.
90
91
  - Sonnet 5 is a conditional route when measured latency, throughput, access,
91
92
  rate limits, or end-to-end workload economics win. Fable 5 is an
92
93
  eligibility-gated specialist; its missing-domain-concept or long-horizon
@@ -95,18 +96,18 @@ publish a provider-independent effort conversion.
95
96
  - CursorBench supports Composer 2.5 as an economical bounded coding worker.
96
97
  Cursor explicitly documents Composer fast as the same intelligence at a
97
98
  higher latency tier price.
98
- - Mechanical reconnaissance remains harness-specific: Luna high in Codex,
99
+ - Mechanical reconnaissance remains harness-specific: GPT-6 Luna high in Codex,
99
100
  Haiku 4.5 in Claude, and Composer 2.5 in Cursor. Direct API mini or nano
100
101
  routes are limited to strict extraction, classification, or similarly
101
102
  mechanically verified work.
102
- - OpenAI's long-context evidence disqualifies Luna for some very large-context
103
- work even when Luna's nominal context window accepts the input. Sol's current
104
- direct-API requests above 272K input tokens carry a 2× input and 1.5× output
105
- price step.
106
- - Sol's measured advantage is trajectory efficiency, not universal list-price
107
- leadership. Opus output token pricing can be lower while Sol costs less per
108
- completed task through fewer tokens, steps, turns, or recoveries.
109
- - Two independent benchmarks show top-end Opus effort is non-monotonic:
103
+ - OpenAI's historical GPT-5.6 long-context evidence cautions against
104
+ assuming Luna is suitable for every large-context workload. Confirm current
105
+ GPT-6 context and pricing in the provider catalogue before quoting a limit
106
+ or price step.
107
+ - Earlier GPT-5.6 Sol trajectory measurements illustrated that list price
108
+ alone does not determine cost per completed task. Re-measure GPT-6 against
109
+ Opus 5.5 before claiming an economic advantage.
110
+ - Historical Opus 5 measurements showed non-monotonic top-end effort:
110
111
  Artificial Analysis Coding Agent Index (xhigh 67, max 66) and Frontier-Bench
111
112
  (xhigh 44.4%, max 43.3%), both retrieved 2026-07-25. Frontier-Bench is
112
113
  vendor-operated and its figure rests on a single self-reported row with no
@@ -116,9 +117,9 @@ publish a provider-independent effort conversion.
116
117
  marginal cost per score point is roughly 3.75x worse above high than below
117
118
  it. Together these support treating max as a route requiring justification,
118
119
  not a strictly better rung.
119
- - Cyber-sensitive and valid dual-use work should start on Opus 5 with
120
- documented fallback handling where safeguards block the workflow. Opus 4.8
121
- remains a compatibility fallback, not the universal primary.
120
+ - Cyber-sensitive and valid dual-use work should start on Opus 5.5 with
121
+ documented fallback handling where safeguards block the workflow. Do not
122
+ route current work to retired Opus 5 or 4.8 by default.
122
123
  - Cursor disclosed that Cursor repository data entered Grok 4.5's training
123
124
  mixture. Its CursorBench ranking is evidence of competitiveness, not proof
124
125
  of superiority over nearby frontier models. Practical use supports Grok 4.5
package/assets/skills/subagent-orchestration/references/provider-claude.md CHANGED
@@ -1,7 +1,7 @@
1
1
  ---
2
- guidance_version: 2026-07-25
3
- last_verified: 2026-07-25
4
- review_after: 2026-09-08
2
+ guidance_version: 2026-09-23
3
+ last_verified: 2026-09-23
4
+ review_after: 2026-12-22
5
5
  ---
6
6
 
7
7
  # Claude Model Selection
@@ -15,15 +15,13 @@ the dated model examples below. Launch mechanics for OAT dispatch live in
15
15
 
16
16
  - `claude-haiku-4-5`: fastest economical class for straightforward,
17
17
  high-volume, mechanically verified work.
18
- - `claude-opus-5`: normal substantive route and the default for
18
+ - `claude-opus-5-5`: normal substantive route and the default for
19
19
  interpretation-heavy implementation, hard reasoning, and consequential work.
20
20
  - `claude-sonnet-5`: conditional route when measured latency, throughput,
21
21
  access, rate limits, or established-workflow economics beat the relevant
22
22
  Opus route.
23
- - `claude-fable-5`: eligibility-gated specialist for a directly relevant
23
+ - `claude-fable-5-1`: eligibility-gated specialist for a directly relevant
24
24
  failure mode, not a universal escalation above Opus.
25
- - `claude-opus-4-8`: compatibility or documented safeguard fallback, not the
26
- current general primary.
27
25
 
28
26
  Claude Mythos 5 is invitation-only and intended for approved defensive cyber
29
27
  work. It is not a general routing candidate unless the organization explicitly
@@ -33,22 +31,31 @@ provisions and authorizes it.
33
31
 
34
32
  Never normalize Claude effort against OpenAI or Cursor labels.
35
33
 
36
- - Opus 5: use medium for normal substantive work and high for hard reasoning,
37
- architecture, ambiguity, and deep review. Use xhigh only when reasoning depth
34
+ For OAT-managed Claude reviewers and phase implementers, explicit effort is
35
+ definition-bound. Classify the task here, choose an eligible configured
36
+ model/effort pair, resolve it, and launch the exact generated agent variant.
37
+ The Agent call has no effort field. If it also supplies a model, that model must
38
+ match the generated definition. Model-only candidates retain their existing
39
+ native model argument behavior, and inherited targets leave both axes to the
40
+ host.
41
+
42
+ - Opus 5.5: use low for bounded intelligent recon with source checks, medium
43
+ for normal substantive work, and high for hard reasoning, architecture,
44
+ ambiguity, and deep review. Use xhigh only when reasoning depth
38
45
  is the bottleneck or an evaluated long-horizon workload benefits. Max is
39
46
  exceptional and requires a workload-specific effort sweep or an explicit
40
47
  quality-first exception.
41
48
  - Sonnet 5: use medium or high only when measured latency, throughput, access,
42
49
  or workload economics justify the conditional route. Do not preserve xhigh
43
50
  or max as generic workhorse settings.
44
- - Fable 5: use high or xhigh only for a qualified specialist case. Max is
51
+ - Fable 5.1: use high or xhigh only for a qualified specialist case. Max is
45
52
  exceptional; medium or low are not routine economy substitutes for Opus.
46
53
  - Haiku 4.5 does not expose the same adaptive-effort surface. Use only controls
47
54
  present in the live schema.
48
55
 
49
56
  Effort changes tool-call behavior as well as prose. Record it independently.
50
- Opus 5 adaptive thinking is on by default; current direct-API behavior rejects
51
- disabling thinking at xhigh or max. Verify the live schema before launch.
57
+ Opus 5.5 adaptive thinking is always on in the current direct API. Verify the
58
+ live schema before launch.
52
59
  Changing effort or speed can invalidate prompt caches. Fable adaptive thinking
53
60
  is always on in the current API.
54
61
 
@@ -58,13 +65,13 @@ These routes are dated guidance and benchmark-derived routing hypotheses.
58
65
  Evaluate them on representative local work before treating fine boundaries as
59
66
  stable.
60
67
 
61
- | Task class | Default | Economy | Escalation | Floor notes |
62
- | ------------------------ | ----------------------------------- | ------------------------------------------------------ | --------------------------------------------------- | ----------------------------------------------------------------------------------------------- |
63
- | `mechanical-recon` | Haiku 4.5 | Same, with strict output and mechanical verification | Opus 5 medium | Haiku is below floor for semantic audits and silent-miss-prone exploration. |
64
- | `intelligent-recon` | Opus 5 medium | Sonnet 5 medium/high only when its measured route wins | Opus 5 high | Do not use Sonnet low for coding or open-ended recon. |
65
- | `default-implementation` | Opus 5 medium | Sonnet 5 medium/high for evaluated bounded throughput | Opus 5 high | Fable is not a routine implementation or economy route. |
66
- | `hard-reasoning` | Opus 5 high | Opus 5 medium only after narrowing and evaluation | Opus 5 xhigh for a reasoning-depth bottleneck | Use max only after a workload-specific effort sweep. |
67
- | `consequential` | Opus 5 high plus independent review | No routine economy route | Opus 5 xhigh when deeper reasoning is also required | Consequence adds review and root authorization; it does not automatically require xhigh or max. |
68
+ | Task class | Default | Economy | Escalation | Floor notes |
69
+ | ------------------------ | ------------------------------------- | ------------------------------------------------------ | ----------------------------------------------------- | ----------------------------------------------------------------------------------------------- |
70
+ | `mechanical-recon` | Haiku 4.5 | Same, with strict output and mechanical verification | Opus 5.5 medium | Haiku is below floor for semantic audits and silent-miss-prone exploration. |
71
+ | `intelligent-recon` | Opus 5.5 low | Sonnet 5 medium/high only when its measured route wins | Opus 5.5 medium; high when deeper reasoning is needed | Verify silent-miss-prone conclusions; do not use Sonnet low for open-ended recon. |
72
+ | `default-implementation` | Opus 5.5 medium | Sonnet 5 medium/high for evaluated bounded throughput | Opus 5.5 high | Fable is not a routine implementation or economy route. |
73
+ | `hard-reasoning` | Opus 5.5 high | Opus 5.5 medium only after narrowing and evaluation | Opus 5.5 xhigh for a reasoning-depth bottleneck | Use max only after a workload-specific effort sweep. |
74
+ | `consequential` | Opus 5.5 high plus independent review | No routine economy route | Opus 5.5 xhigh when deeper reasoning is also required | Consequence adds review and root authorization; it does not automatically require xhigh or max. |
68
75
 
69
76
  ## Root and Subagent Cost Posture
70
77
 
@@ -80,7 +87,7 @@ consequence, or a directly relevant Fable strength exposes a failure mode that
80
87
  Fable is expected to catch and eligibility permits it. This named Fable
81
88
  instantiation is provisional.
82
89
 
83
- A consequential classification by itself is insufficient.
90
+ A stronger safety classifier is a specialist consideration, not an exception that inverts the general Opus-first policy. A consequential classification by itself is insufficient.
84
91
 
85
92
  ## Conditional and Specialist Routes
86
93
 
@@ -101,16 +108,17 @@ verified before selection.
101
108
 
102
109
  Eligibility filters the candidate set before capability or economy is compared.
103
110
  A route that fails one of these gates is unavailable regardless of benchmark
104
- standing. These are dated facts verified 2026-07-25; reverify against the live
105
- schema and the organization's actual plan before launch.
106
-
107
- | Constraint | Opus 5 | Sonnet 5 | Fable 5 | Haiku 4.5 |
108
- | ------------------- | ---------------------- | ------------- | -------------------------- | ------------- |
109
- | Zero data retention | eligible | eligible | **unavailable**, needs 30d | eligible |
110
- | Priority Tier | not supported | not supported | supported | verify live |
111
- | Fast mode | gated research preview | no | no | no |
112
- | Long context | 1M at standard rates | 1M standard | 1M standard | verify live |
113
- | Claude Code minimum | version-gated | version-gated | version-gated | version-gated |
111
+ standing. The non-Opus constraints are dated observations; reverify against the live
112
+ schema and the organization's actual plan before launch. Do not transfer Opus 5
113
+ service-tier or retention claims to Opus 5.5 without fresh evidence.
114
+
115
+ | Constraint | Opus 5.5 | Sonnet 5 | Fable 5.1 | Haiku 4.5 |
116
+ | ------------------- | ------------- | ------------- | -------------------------- | ------------- |
117
+ | Zero data retention | verify live | eligible | **unavailable**, needs 30d | eligible |
118
+ | Priority Tier | verify live | not supported | supported | verify live |
119
+ | Fast mode | verify live | no | no | no |
120
+ | Long context | verify live | 1M standard | 1M standard | verify live |
121
+ | Claude Code minimum | version-gated | version-gated | version-gated | version-gated |
114
122
 
115
123
  - Zero data retention removes Fable from the route set outright. Check the
116
124
  retention requirement before selecting a specialist reviewer.
@@ -129,13 +137,10 @@ from `model-selection-principles.md`.
129
137
  ## Cyber-Sensitive Evidence
130
138
 
131
139
  For security review, vulnerability triage, auth boundaries, permissions, and
132
- other valid dual-use workflows, start with Opus 5 at the effort warranted by
140
+ other valid dual-use workflows, start with Opus 5.5 at the effort warranted by
133
141
  reasoning depth and pair consequential findings with an independent provider
134
142
  review.
135
143
 
136
144
  If a safeguard blocks a valid workflow, record the refusal, preserve the
137
- authorization boundary, and use a documented compatible fallback. Opus 4.8 is
138
- the current fallback target where necessary, not the universal cyber primary.
139
- Do not infer that a stronger safety classifier is a capability weakness. It is
140
- evidence about predictable workflow completion and refusal behavior, not an
141
- exception that inverts the general Opus-first policy.
145
+ authorization boundary, and choose only a currently supported, authorized
146
+ compatible route. Retired Opus 4.8/5.0 generations are not active fallbacks.
package/assets/skills/subagent-orchestration/references/provider-codex.md CHANGED
@@ -1,7 +1,7 @@
1
1
  ---
2
- guidance_version: 2026-07-25
3
- last_verified: 2026-07-25
4
- review_after: 2026-09-08
2
+ guidance_version: 2026-09-22
3
+ last_verified: 2026-09-22
4
+ review_after: 2026-12-22
5
5
  ---
6
6
 
7
7
  # Codex and OpenAI Model Selection
@@ -13,12 +13,14 @@ model examples below. Launch mechanics for OAT dispatch live in
13
13
 
14
14
  ## Current Families
15
15
 
16
- Use the GPT-5.6 family for new general coding and knowledge work unless an
17
- existing evaluated workflow requires an older snapshot:
16
+ Use the verified GPT-6 Sol/Luna model IDs for new Codex work:
18
17
 
19
- - `gpt-5.6-sol`: frontier capability;
20
- - `gpt-5.6-terra`: intelligence and cost balance;
21
- - `gpt-5.6-luna`: cost-sensitive, high-volume work.
18
+ - `gpt-6-sol`: high-capability implementation and reasoning route;
19
+ - `gpt-6-luna`: cost-sensitive, bounded work.
20
+
21
+ The GPT-5.6 Sol/Luna/Terra targets remain supported for explicit configurations
22
+ and established workloads. Re-evaluate economics and qualitative routing on
23
+ representative work before carrying over GPT-5.6-specific comparisons.
22
24
 
23
25
  Direct API specialist routes:
24
26
 
@@ -33,36 +35,29 @@ name includes `codex`.
33
35
 
34
36
  ## Dated Task-Class Matrix
35
37
 
36
- | Task class | Default | Economy | Escalation | Floor notes |
37
- | ------------------------ | ---------------------------------------------- | -------------------------------------------------------------------------------------------------- | ---------------------------------------------------------------------------------------------------------------- | ----------------------------------------------------------------------------------------------- |
38
- | `mechanical-recon` | `gpt-5.6-luna`, `high` | Direct API only: `gpt-5.4-mini` medium; `gpt-5.4-nano` medium for strict extraction/classification | `gpt-5.6-terra`, medium or high | Do not use Luna none/low for broad repository work. |
39
- | `intelligent-recon` | `gpt-5.6-terra`, `high` | Terra `medium` with tight scope and verification | Sol `medium`, then `high` | A silent miss disqualifies Terra none/low. |
40
- | `default-implementation` | `gpt-5.6-sol`, `medium` | Terra `high` for independently bounded, strongly tested work | Sol `high` | Luna is not the normal implementation lead. |
41
- | `hard-reasoning` | `gpt-5.6-sol`, `high` | Sol `medium` only after narrowing the problem | Sol `xhigh` | Do not automatically escalate Terra to xhigh/max; move to Sol. |
42
- | `consequential` | `gpt-5.6-sol`, `high`, plus independent review | No routine economy route | Sol `xhigh` when deeper reasoning is required; `max` or a separately evaluated `pro` route only after evaluation | Consequence adds review and root authorization; it does not automatically require xhigh or max. |
43
-
44
- `reasoning.mode: "pro"` is a separate quality/latency/cost control on GPT-5.6,
45
- not a model slug and not an effort label. Record it independently. Evaluate it
46
- against max or xhigh on representative consequential work before adopting it.
47
-
48
- The task-class ladder is work-shape based. Sol is the code-first,
49
- trajectory-efficient route for implementation and hard reasoning; this does
50
- not establish a provider-independent ranking against interpretation-heavy
51
- models in other harnesses. Consequential work requires independent review and
52
- root-owned authorization. It does not automatically force Sol xhigh or max:
53
- choose effort from reasoning depth, then add the consequence controls.
38
+ These are current task-class starting points, not an evaluated ordering across
39
+ providers or a promise about relative GPT-6 latency and price. All shown
40
+ model/effort pairs are in the local Codex catalogue.
54
41
 
55
- ## Long-Context Floor
42
+ | Task class | Default | Escalation | Floor notes |
43
+ | ------------------------ | ------------------------------------------ | --------------------------------------------------------------------- | ----------------------------------------------------------------------------- |
44
+ | `mechanical-recon` | `gpt-6-luna`, high | `gpt-6-sol`, medium | Lower effort is suitable only when a miss is cheap and mechanically detected. |
45
+ | `intelligent-recon` | `gpt-6-sol`, medium | Sol high | Preserve verification for silent-miss-prone reconnaissance. |
46
+ | `default-implementation` | `gpt-6-sol`, medium | Sol high | Luna is for independently bounded, strongly verified work. |
47
+ | `hard-reasoning` | `gpt-6-sol`, high | Sol xhigh | Narrow the problem before raising effort. |
48
+ | `consequential` | `gpt-6-sol`, high, plus independent review | Sol xhigh for a reasoning-depth bottleneck; max only after evaluation | The root retains consequential authorization. |
56
49
 
57
- For very large context, prefer Sol or Terra. Published GPT-5.6 results show a
58
- large Luna retention drop in the 256K-to-1M range. Large context does not change
59
- the task class, but it may disqualify Luna, mini, or nano.
50
+ The bundled frontier ladder and the supported Sol effort set end at max.
51
+ GPT-5.6 `reasoning.mode: "pro"` is a
52
+ separate quality/latency/cost control on that older model, not a GPT-6 effort
53
+ label or model slug.
54
+
55
+ ## Long-Context Floor
60
56
 
61
- For Sol requests above 272K input tokens, the current direct-API price step is
62
- 2× input and 1.5× output. Apply the threshold to the whole request and verify
63
- the live pricing contract before cost-sensitive dispatch. Crossing this
64
- threshold can change the economical route even when Sol remains the capability
65
- choice.
57
+ The former 256K-to-1M Luna retention observation and Sol 272K price step
58
+ pertain to GPT-5.6. Do not extrapolate either to GPT-6. Verify GPT-6 context
59
+ and pricing contracts against current official documentation and the actual
60
+ runtime before cost-sensitive or very large-context dispatch.
66
61
 
67
62
  ## Trajectory Economics
68
63