@open-agent-toolkit/cli 0.3.0 → 0.3.2

package/assets/skills/subagent-orchestration/references/provider-cursor.md

@@ -1,11 +1,10 @@
 ---
-guidance_version: 2026-07-25
-last_verified: 2026-07-25
-review_after: 2026-09-04
+guidance_version: 2026-09-23
+last_verified: 2026-09-23
+review_after: 2026-12-22
 catalog_basis:
-  live `cursor-agent models` catalog and CursorBench 3.2, with Opus 5
-  and Opus 4.8 pin selectors probe-verified against Cursor 3.12.30 via
-  subagentStart lifecycle hooks
+  Cursor desktop 3.20.14 resolved all five Opus 5.5 effort selectors in native
+  subagent hooks; GPT-6 Sol/Luna were absent from the observed Cursor catalog
 ---
 
 # Cursor Model Selection
@@ -45,25 +44,40 @@ does not by itself prove lower user-observed elapsed time.
 
 ## Dated Task-Class Matrix
 
-Use exact aliases from the live catalog. The examples below come from two
-observations, not one. Most were present in a 2026-07-21 catalog snapshot. The
-`claude-opus-5` and `claude-opus-4-8` routes postdate it — Opus 5 released
-2026-07-24 — and were confirmed against the live catalog and probe-verified on
-2026-07-25, as recorded under Verified Opus 5 Cursor Routes below. The
-remainder of the catalog has not been re-snapshotted since 2026-07-21, so treat
-non-Opus entries as the older and weaker observation.
-
-| Task class               | Default                                                                                                                                                      | Economy                                                                              | Escalation                                                                                                             | Floor notes                                                                                                                                                                                                                                                                     |
-| ------------------------ | ------------------------------------------------------------------------------------------------------------------------------------------------------------ | ------------------------------------------------------------------------------------ | ---------------------------------------------------------------------------------------------------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
-| `mechanical-recon`       | `composer-2.5`                                                                                                                                               | `composer-2.5`                                                                       | `gpt-5.6-luna-high`                                                                                                    | Use `composer-2.5-fast` or Luna fast only for latency, not capability.                                                                                                                                                                                                          |
-| `intelligent-recon`      | `cursor-grok-4.5-medium` or `gpt-5.6-sol-medium`                                                                                                             | `gpt-5.6-terra-high`                                                                 | `gpt-5.6-sol-high`                                                                                                     | Grok medium is a primary route; Sol medium is the conservative silent-miss alternative; Terra high is the cost-saving route.                                                                                                                                                    |
-| `default-implementation` | `gpt-5.6-sol-medium`                                                                                                                                         | `composer-2.5` for bounded, testable changes                                         | `claude-opus-5-thinking-medium`                                                                                        | Sol is the code-first incumbent; Opus 5 medium is the substantive route for interpretation-heavy work; `cursor-grok-4.5-medium` remains a primary alternative.                                                                                                                  |
-| `hard-reasoning`         | `claude-opus-5-thinking-high`                                                                                                                                | `cursor-grok-4.5-high` for architecture, ambiguous debugging, and incident diagnosis | `claude-opus-5-thinking-xhigh`, then `gpt-5.6-sol-xhigh` for code-first work                                           | Opus 5 high is the starting point. Escalate to xhigh only for a reasoning-depth bottleneck. Do not infer cross-provider effort equivalence.                                                                                                                                     |
-| `consequential`          | The task's own reasoning-depth route, plus independent cross-family review: typically `gpt-5.6-sol-high` authoring with `claude-opus-5-thinking-high` review | No routine economy route                                                             | Raise the pair to xhigh for a reasoning-depth bottleneck; Sol max plus Opus 5 max only after a workload-specific sweep | Consequence adds independent cross-family review and root authorization, not automatic top effort. Select effort from reasoning depth exactly as the non-consequential row would, then add independence on top. For cyber-sensitive review, do not rely on Fable or Grok alone. |
-
-## Verified Opus 5 Cursor Routes
-
-Opus 5 is a qualified Cursor route. Five effort rungs plus an Opus 4.8 rung were
+Use currently approved exact aliases from the Cursor catalogue. On 2026-09-23,
+Cursor desktop 3.20.14 resolved all five `claude-opus-5-5[effort=...]`
+selectors to corresponding flat IDs in native hooks. See the
+probe record at `packages/cli/src/providers/cursor/codec/__fixtures__/README.md`.
+GPT-6 Sol/Luna were absent from the observed Cursor catalogue, so their Codex
+availability does not authorize Cursor routes.
+
+| Task class               | Default                                                             | Economy                                            | Escalation                                                  | Floor notes                                                                                                      |
+| ------------------------ | ------------------------------------------------------------------- | -------------------------------------------------- | ----------------------------------------------------------- | ---------------------------------------------------------------------------------------------------------------- |
+| `mechanical-recon`       | `composer-2.5`                                                      | Same with mechanical verification                  | `gpt-5.6-luna-high`                                         | Fast aliases select service tier, not capability.                                                                |
+| `intelligent-recon`      | `cursor-grok-4.5-medium` or `gpt-5.6-sol-medium`                    | `gpt-5.6-terra-high`                               | `gpt-5.6-sol-high`                                          | Verify silent-miss-prone conclusions.                                                                            |
+| `default-implementation` | `gpt-5.6-sol-medium`                                                | `composer-2.5` for bounded, testable changes       | `gpt-5.6-sol-high`                                          | `cursor-grok-4.5-medium` remains an available judgment-heavy alternative.                                        |
+| `hard-reasoning`         | `gpt-5.6-sol-high`                                                  | `cursor-grok-4.5-high` with corroborating evidence | `gpt-5.6-sol-xhigh`                                         | Opus 5.5 high/xhigh are verified cross-family alternatives; qualify task performance separately.                 |
+| `consequential`          | An eligible high-effort author with independent cross-family review | No routine economy route                           | Sol xhigh for a depth bottleneck; max only after evaluation | If independent review cannot be met among approved Cursor mappings, use a separately qualified provider or stop. |
+
+## Current Opus 5.5 Pin Evidence
+
+Cursor desktop 3.20.14 resolved `claude-opus-5-5[effort=low|medium|high|xhigh|max]`
+to `claude-opus-5-5-low|medium|high|xhigh|max`, respectively, on 2026-09-23.
+Both the `subagentStart.subagent_model` and the subagent Shell `preToolUse.model`
+agreed for every rung. The Sonnet 5 high positive control resolved to
+`claude-sonnet-5-thinking-high`. An unknown family fell back to
+`cursor-grok-4.6-high-fast`; an unknown effort on Opus 5.5 fell back to
+`claude-opus-5-5-medium`. The provenance summary at `packages/cli/src/providers/cursor/codec/__fixtures__/README.md`
+links the adjacent redacted summary and native-event JSONL records containing
+the exact observations. Recheck live availability before launch.
+
+## Historical Opus Cursor Probe Evidence (Retired)
+
+The following observations document Cursor 3.12.30 behavior on 2026-07-25.
+They no longer qualify those generations for current recommendations or the
+active approved mapping catalogue. A new family needs its own live probe.
+
+Five Opus 5 effort rungs plus an Opus 4.8 rung were
 probe-verified on 2026-07-25 against Cursor 3.12.30, using `subagentStart`
 lifecycle hooks to read the resolved model that Cursor itself reports:
 
@@ -80,21 +94,15 @@ Every rung resolves to the thinking variant. This is a verified fact, not an
 inference: the catalog carries distinct non-thinking IDs at low, medium, and
 high, and none were selected.
 
-Routing follows the accepted conclusions rather than the effort ladder alone.
-Opus 5 medium is the normal substantive route, high is the hard-reasoning
-starting point, xhigh is selective for a reasoning-depth bottleneck or an
-evaluated long-horizon benefit, and max requires a workload-specific sweep or an
-explicit exception. Sol remains the code-first, trajectory-efficient engineering
-route; Opus is generally stronger for interpretation-heavy work. Exact
-cross-provider substitutions remain provisional.
-
-Opus 4.8 xhigh is catalogued so the cyber-sensitive route is dispatchable, but it
-is deliberately excluded from the bundled recommendation. Select it explicitly
-when the cyber-sensitive refusal profile is required.
+That historical result did not validate an Opus 5.5 selector; the 2026-09-23
+probe above independently qualifies its five selectors. Even where the
+requested effort has the same spelling across generations, preserve the old
+probe only as evidence of past behavior and use the current approved catalogue
+for dispatch.
 
 ### Unresolvable selectors fall back to a default, silently
 
-Probing showed that Cursor does not reject a malformed pin. It substitutes a
+The historical probe showed that Cursor did not reject a malformed pin. It substitutes a
 default for whichever component it cannot resolve, with no error or warning:
 
 - An unknown family falls back to the default model. `claude-opus-9[effort=high]`
@@ -102,7 +110,7 @@ default for whichever component it cannot resolve, with no error or warning:
 - An unknown effort falls back to the family's default rung.
   `claude-opus-5[effort=ultra]` resolved to `claude-opus-5-thinking-high`.
 
-The default rung is family-specific, not a fixed value. In the live catalog
+The default rung is family-specific, not a fixed value. In that catalog snapshot
 exactly one rung per family carries an unqualified display label, and that marks
 the default: `claude-opus-5-high` shows as `Opus 5 1M` while
 `claude-opus-5-medium` shows as `Opus 5 1M Medium`. The default is not always

package/assets/templates/plan.md

@@ -55,13 +55,13 @@ Default is `[]` (fully sequential, no worktrees). Only declare parallelism when
 
 _Optional override surface. Use only for explicit user-authored constraints or preferences. Omit this section when runtime selection should choose the lowest confident tier._
 
-Blank or `auto` means there is no explicit constraint for that provider. Do not generate rows by default; a missing phase row uses runtime selection.
+Blank or `auto` means there is no explicit phase constraint. Do not generate rows by default; a missing phase row uses runtime selection.
 
-| Phase | Claude model                     | Codex effort                   | Rationale                     |
-| ----- | -------------------------------- | ------------------------------ | ----------------------------- |
-| pNN   | haiku\|sonnet\|opus\|fable\|auto | low\|medium\|high\|xhigh\|auto | why this constraint is needed |
+| Phase | Named ceiling                           | Rationale                     |
+| ----- | --------------------------------------- | ----------------------------- |
+| pNN   | economy\|balanced\|high\|frontier\|auto | why this constraint is needed |
 
-Codex effort values are preferred controls. `oat-project-implement` caps them when a capped managed dispatch policy exists, selects them directly under managed `Uncapped`, and maps selected efforts to pinned implementer variants when available. Codex provider default effort is informational only for explicit inherit/default behavior or base/unpinned fallback paths.
+The named ceiling is a maximum candidate tier, not an exact provider, model family, or effort preference. `oat-project-implement` chooses one exact phase-implementer target from the complete configured ladder at runtime.
 
 ---
 

package/dist/commands/gate/review-verdict.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"review-verdict.d.ts","sourceRoot":"","sources":["../../../src/commands/gate/review-verdict.ts"],"names":[],"mappings":"AAMA,MAAM,WAAW,iBAAiB;IAChC,YAAY,EAAE,MAAM,CAAC;IACrB,UAAU,EAAE,MAAM,GAAG,UAAU,GAAG,SAAS,CAAC;IAC5C,KAAK,EAAE,MAAM,GAAG,IAAI,CAAC;IACrB,UAAU,EAAE,MAAM,GAAG,IAAI,CAAC;IAC1B,OAAO,EAAE,MAAM,GAAG,IAAI,CAAC;IACvB,cAAc,CAAC,EAAE,4BAA4B,CAAC;IAC9C,MAAM,EAAE;QACN,QAAQ,EAAE,MAAM,CAAC;QACjB,IAAI,EAAE,MAAM,CAAC;QACb,MAAM,EAAE,MAAM,CAAC;QACf,GAAG,EAAE,MAAM,CAAC;KACb,CAAC;IACF,QAAQ,EAAE,OAAO,CAAC;IAClB,aAAa,CAAC,EAAE;QACd,kBAAkB,EAAE,QAAQ,EAAE,CAAC;QAC/B,SAAS,EAAE,OAAO,CAAC;KACpB,CAAC;CACH;AAED,MAAM,WAAW,4BAA4B;IAC3C,KAAK,EAAE,MAAM,GAAG,IAAI,CAAC;IACrB,QAAQ,EAAE,MAAM,GAAG,IAAI,CAAC;IACxB,OAAO,EAAE,MAAM,GAAG,IAAI,CAAC;IACvB,KAAK,EAAE,MAAM,GAAG,IAAI,CAAC;IACrB,eAAe,EAAE,MAAM,GAAG,IAAI,CAAC;IAC/B,MAAM,EAAE,MAAM,GAAG,IAAI,CAAC;CACvB;AAED,MAAM,MAAM,QAAQ,GAAG,MAAM,iBAAiB,CAAC,QAAQ,CAAC,CAAC;AAEzD,MAAM,WAAW,0BAA0B;IACzC,QAAQ,CAAC,OAAO,EAAE,MAAM,CAAC;IACzB,QAAQ,CAAC,SAAS,EAAE,MAAM,CAAC;CAC5B;AAED,MAAM,WAAW,6BAA6B;IAC5C,qCAAqC,CAAC,EAAE,OAAO,CAAC;IAChD,gBAAgB,CAAC,EAAE,0BAA0B,CAAC;CAC/C;AAmOD,wBAAgB,mBAAmB,CAAC,QAAQ,EAAE,QAAQ,GAAG,MAAM,CAE9D;AA2eD,wBAAsB,sBAAsB,CAC1C,YAAY,EAAE,MAAM,EACpB,OAAO,GAAE,6BAAkC,GAC1C,OAAO,CAAC,iBAAiB,CAAC,CA8F5B"}
+{"version":3,"file":"review-verdict.d.ts","sourceRoot":"","sources":["../../../src/commands/gate/review-verdict.ts"],"names":[],"mappings":"AAMA,MAAM,WAAW,iBAAiB;IAChC,YAAY,EAAE,MAAM,CAAC;IACrB,UAAU,EAAE,MAAM,GAAG,UAAU,GAAG,SAAS,CAAC;IAC5C,KAAK,EAAE,MAAM,GAAG,IAAI,CAAC;IACrB,UAAU,EAAE,MAAM,GAAG,IAAI,CAAC;IAC1B,OAAO,EAAE,MAAM,GAAG,IAAI,CAAC;IACvB,cAAc,CAAC,EAAE,4BAA4B,CAAC;IAC9C,MAAM,EAAE;QACN,QAAQ,EAAE,MAAM,CAAC;QACjB,IAAI,EAAE,MAAM,CAAC;QACb,MAAM,EAAE,MAAM,CAAC;QACf,GAAG,EAAE,MAAM,CAAC;KACb,CAAC;IACF,QAAQ,EAAE,OAAO,CAAC;IAClB,aAAa,CAAC,EAAE;QACd,kBAAkB,EAAE,QAAQ,EAAE,CAAC;QAC/B,SAAS,EAAE,OAAO,CAAC;KACpB,CAAC;CACH;AAED,MAAM,WAAW,4BAA4B;IAC3C,KAAK,EAAE,MAAM,GAAG,IAAI,CAAC;IACrB,QAAQ,EAAE,MAAM,GAAG,IAAI,CAAC;IACxB,OAAO,EAAE,MAAM,GAAG,IAAI,CAAC;IACvB,KAAK,EAAE,MAAM,GAAG,IAAI,CAAC;IACrB,eAAe,EAAE,MAAM,GAAG,IAAI,CAAC;IAC/B,MAAM,EAAE,MAAM,GAAG,IAAI,CAAC;CACvB;AAED,MAAM,MAAM,QAAQ,GAAG,MAAM,iBAAiB,CAAC,QAAQ,CAAC,CAAC;AAEzD,MAAM,WAAW,0BAA0B;IACzC,QAAQ,CAAC,OAAO,EAAE,MAAM,CAAC;IACzB,QAAQ,CAAC,SAAS,EAAE,MAAM,CAAC;CAC5B;AAED,MAAM,WAAW,6BAA6B;IAC5C,qCAAqC,CAAC,EAAE,OAAO,CAAC;IAChD,gBAAgB,CAAC,EAAE,0BAA0B,CAAC;CAC/C;AA8OD,wBAAgB,mBAAmB,CAAC,QAAQ,EAAE,QAAQ,GAAG,MAAM,CAE9D;AA4fD,wBAAsB,sBAAsB,CAC1C,YAAY,EAAE,MAAM,EACpB,OAAO,GAAE,6BAAkC,GAC1C,OAAO,CAAC,iBAAiB,CAAC,CA8F5B"}

package/dist/commands/gate/review-verdict.js

@@ -129,6 +129,17 @@ function sectionContentIsEmpty(content) {
         .trim();
     return cleaned.length === 0 || /^none\.?$/i.test(cleaned);
 }
+/**
+ * Count the findings a severity section declares.
+ *
+ * Only top-level markdown list items count. Prose shapes are deliberately not
+ * counted — a bold lead-in paragraph, a wrapped continuation line, and a nested
+ * detail bullet are indistinguishable from a finding title without parsing the
+ * prose, so counting them would miscount and double-count. The cost of that
+ * choice is that a reviewer who writes findings as paragraphs gets a zero
+ * tally, so every diagnostic built on this count must name the list-item
+ * requirement (see `FINDING_LIST_ITEM_CONTRACT_HINT`).
+ */
 function countFindingsInSection(content) {
     if (sectionContentIsEmpty(content)) {
         return 0;
@@ -409,6 +420,15 @@ async function normalizeMissingEmptySeveritySections(artifactPath, content, coun
         persisted: insertedSeverities.length > 0 && persist,
     };
 }
+/**
+ * The one fact a reviewer cannot recover from the raw contradiction report.
+ *
+ * A section whose findings are written as bold paragraphs tallies zero, so the
+ * gate reports a contradiction that looks like a counting mistake. Without this
+ * sentence the reviewer has no way to learn that the shape of the finding, not
+ * the arithmetic, is what the parser objected to.
+ */
+const FINDING_LIST_ITEM_CONTRACT_HINT = 'Only markdown list items count as findings in a Findings section: a line starting with `- `, `* `, `+ `, or `1. `. A finding written as a bold paragraph, a heading, or any other prose shape counts as zero, which is the usual reason a section tallies lower than the declared counts. Rewrite each finding as a list item, or correct the declared counts to match.';
 function formatCounts(counts) {
     return SEVERITIES.map((severity) => `${counts[severity]} ${severity}`).join(', ');
 }
@@ -424,7 +444,8 @@ function assertAgreeingCounts(artifactPath, left, right, tiers = SEVERITIES) {
     if (differing.length === 0) {
         return;
     }
-    throw new Error(`Review artifact at ${artifactPath} contradicts itself about finding counts (${differing.join(', ')}): declared counts are ${formatCounts(left.counts)} (${left.name}) but ${formatCounts(right.counts)} (${right.name}). Reconcile them before the gate can evaluate the review.`);
+    const sectionsInvolved = left.fromSections === true || right.fromSections === true;
+    throw new Error(`Review artifact at ${artifactPath} contradicts itself about finding counts (${differing.join(', ')}): declared counts are ${formatCounts(left.counts)} (${left.name}) but ${formatCounts(right.counts)} (${right.name}). Reconcile them before the gate can evaluate the review.${sectionsInvolved ? ` ${FINDING_LIST_ITEM_CONTRACT_HINT}` : ''}`);
 }
 /**
  * Resolve the artifact's counts, cross-checking every source that is present.
@@ -467,6 +488,7 @@ function resolveCounts(content, frontmatter, artifactPath) {
             name: 'the Findings sections',
             counts: bodyCounts,
             tiers: bodyTiers,
+            fromSections: true,
         };
         for (const source of explicitSources) {
             const comparableTiers = source.tiers.filter((severity) => bodyTiers.includes(severity));

package/dist/commands/project/dispatch/record.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"record.d.ts","sourceRoot":"","sources":["../../../../src/commands/project/dispatch/record.ts"],"names":[],"mappings":"AAcA,OAAO,EAKL,KAAK,qBAAqB,EAC3B,MAAM,6CAA6C,CAAC;AACrD,OAAO,EAOL,KAAK,wBAAwB,EAE7B,KAAK,4BAA4B,EAElC,MAAM,yCAAyC,CAAC;AAWjD;;;;;GAKG;AACH,MAAM,WAAW,0BAA0B;IACzC,SAAS,CAAC,EAAE,MAAM,OAAO,CAAC,IAAI,CAAC,CAAC;CACjC;AAoCD;;;;;;;;;GASG;AAEH,wBAAgB,qBAAqB,CACnC,OAAO,EAAE,MAAM,EACf,KAAK,GAAE;IACL,OAAO,CAAC,EAAE,MAAM,GAAG,IAAI,CAAC;IACxB,IAAI,CAAC,EAAE,MAAM,GAAG,IAAI,CAAC;IACrB,IAAI,CAAC,EAAE,MAAM,GAAG,IAAI,CAAC;CACjB,GACL,MAAM,CAcR;AAaD,MAAM,WAAW,mBAAmB;IAClC,MAAM,EAAE,qBAAqB,CAAC;IAC9B,KAAK,EAAE,wBAAwB,CAAC;IAChC,uEAAuE;IACvE,iBAAiB,CAAC,EAAE,MAAM,GAAG,IAAI,CAAC;CACnC;AAED;;;;;GAKG;AACH,MAAM,WAAW,6BAA6B;IAC5C,UAAU,EAAE;QACV,QAAQ,EAAE,MAAM,CAAC;QACjB,YAAY,EAAE,MAAM,GAAG,IAAI,CAAC;QAC5B,KAAK,EAAE,MAAM,GAAG,IAAI,CAAC;QACrB,MAAM,EAAE,MAAM,GAAG,IAAI,CAAC;QACtB,WAAW,EAAE,MAAM,GAAG,IAAI,CAAC;KAC5B,CAAC;IACF,QAAQ,EAAE;QACR,QAAQ,EAAE,MAAM,CAAC;QACjB,MAAM,EAAE,MAAM,CAAC;QACf,UAAU,EAAE,MAAM,CAAC;QACnB,YAAY,EAAE,MAAM,GAAG,IAAI,CAAC;QAC5B,IAAI,EAAE,MAAM,GAAG,IAAI,CAAC;QACpB,KAAK,EAAE,MAAM,GAAG,IAAI,CAAC;QACrB,MAAM,EAAE,MAAM,GAAG,IAAI,CAAC;QACtB,WAAW,EAAE,MAAM,GAAG,IAAI,CAAC;KAC5B,GAAG,IAAI,CAAC;IACT,KAAK,EAAE,UAAU,GAAG,aAAa,GAAG,gBAAgB,GAAG,IAAI,CAAC;IAC5D;;;OAGG;IACH,YAAY,EAAE,SAAS,MAAM,EAAE,CAAC;IAChC;;;;OAIG;IACH,MAAM,EAAE,MAAM,GAAG,IAAI,CAAC;IACtB,MAAM,EAAE,UAAU,GAAG,cAAc,CAAC;CACrC;AAED,MAAM,MAAM,2BAA2B,GAAG;IACxC,MAAM,EAAE,WAAW,GAAG,gBAAgB,CAAC;IACvC,IAAI,EAAE,MAAM,GAAG,IAAI,CAAC;IACpB,OAAO,EAAE,OAAO,CAAC;IACjB,MAAM,EAAE,4BAA4B,CAAC;IACrC,eAAe,EAAE,6BAA6B,CAAC;CAChD,CAAC;AAsEF,wBAAgB,wBAAwB,CAAC,KAAK,EAAE,OAAO,GAAG,mBAAmB,CAgF5E;AAgKD;;;;;;GAMG;AACH,wBAAsB,qBAAqB,CAAC,KAAK,EAAE;IACjD,WAAW,EAAE,MAAM,GAAG,IAAI,CAAC;IAC3B,KAAK,EAAE,OAAO,CAAC;IACf,YAAY,CAAC,EAAE,0BAA0B,CAAC;CAC3C,GAAG,OAAO,CAAC,2BAA2B,CAAC,CA6HvC"}
+{"version":3,"file":"record.d.ts","sourceRoot":"","sources":["../../../../src/commands/project/dispatch/record.ts"],"names":[],"mappings":"AAkBA,OAAO,EAKL,KAAK,qBAAqB,EAC3B,MAAM,6CAA6C,CAAC;AACrD,OAAO,EAOL,KAAK,wBAAwB,EAE7B,KAAK,4BAA4B,EAElC,MAAM,yCAAyC,CAAC;AAWjD;;;;;GAKG;AACH,MAAM,WAAW,0BAA0B;IACzC,SAAS,CAAC,EAAE,MAAM,OAAO,CAAC,IAAI,CAAC,CAAC;CACjC;AAoCD;;;;;;;;;GASG;AAEH,wBAAgB,qBAAqB,CACnC,OAAO,EAAE,MAAM,EACf,KAAK,GAAE;IACL,OAAO,CAAC,EAAE,MAAM,GAAG,IAAI,CAAC;IACxB,IAAI,CAAC,EAAE,MAAM,GAAG,IAAI,CAAC;IACrB,IAAI,CAAC,EAAE,MAAM,GAAG,IAAI,CAAC;CACjB,GACL,MAAM,CAcR;AAaD,MAAM,WAAW,mBAAmB;IAClC,MAAM,EAAE,qBAAqB,CAAC;IAC9B,KAAK,EAAE,wBAAwB,CAAC;IAChC,uEAAuE;IACvE,iBAAiB,CAAC,EAAE,MAAM,GAAG,IAAI,CAAC;CACnC;AAED;;;;;GAKG;AACH,MAAM,WAAW,6BAA6B;IAC5C,UAAU,EAAE;QACV,QAAQ,EAAE,MAAM,CAAC;QACjB,YAAY,EAAE,MAAM,GAAG,IAAI,CAAC;QAC5B,KAAK,EAAE,MAAM,GAAG,IAAI,CAAC;QACrB,MAAM,EAAE,MAAM,GAAG,IAAI,CAAC;QACtB,WAAW,EAAE,MAAM,GAAG,IAAI,CAAC;KAC5B,CAAC;IACF,QAAQ,EAAE;QACR,QAAQ,EAAE,MAAM,CAAC;QACjB,MAAM,EAAE,MAAM,CAAC;QACf,UAAU,EAAE,MAAM,CAAC;QACnB,YAAY,EAAE,MAAM,GAAG,IAAI,CAAC;QAC5B,IAAI,EAAE,MAAM,GAAG,IAAI,CAAC;QACpB,KAAK,EAAE,MAAM,GAAG,IAAI,CAAC;QACrB,MAAM,EAAE,MAAM,GAAG,IAAI,CAAC;QACtB,WAAW,EAAE,MAAM,GAAG,IAAI,CAAC;KAC5B,GAAG,IAAI,CAAC;IACT,KAAK,EAAE,UAAU,GAAG,aAAa,GAAG,gBAAgB,GAAG,IAAI,CAAC;IAC5D;;;OAGG;IACH,YAAY,EAAE,SAAS,MAAM,EAAE,CAAC;IAChC;;;;OAIG;IACH,MAAM,EAAE,MAAM,GAAG,IAAI,CAAC;IACtB,MAAM,EAAE,UAAU,GAAG,cAAc,CAAC;CACrC;AAED,MAAM,MAAM,2BAA2B,GAAG;IACxC,MAAM,EAAE,WAAW,GAAG,gBAAgB,CAAC;IACvC,IAAI,EAAE,MAAM,GAAG,IAAI,CAAC;IACpB,OAAO,EAAE,OAAO,CAAC;IACjB,MAAM,EAAE,4BAA4B,CAAC;IACrC,eAAe,EAAE,6BAA6B,CAAC;CAChD,CAAC;AAsEF,wBAAgB,wBAAwB,CAAC,KAAK,EAAE,OAAO,GAAG,mBAAmB,CA0G5E;AAgKD;;;;;;GAMG;AACH,wBAAsB,qBAAqB,CAAC,KAAK,EAAE;IACjD,WAAW,EAAE,MAAM,GAAG,IAAI,CAAC;IAC3B,KAAK,EAAE,OAAO,CAAC;IACf,YAAY,CAAC,EAAE,0BAA0B,CAAC;CAC3C,GAAG,OAAO,CAAC,2BAA2B,CAAC,CA6HvC"}

package/dist/commands/project/dispatch/record.js

@@ -1,6 +1,7 @@
 import { readFile, readdir } from 'node:fs/promises';
 import { join } from 'node:path';
 import { publishContainedJsonRevision, redactedFsError, withContainedWriterLock, } from '../../../fs/io.js';
+import { acceptClaudeLaunchEnvelope, buildClaudeDispatchRecord, } from '../../../providers/claude/dispatch-envelope.js';
 import { assertJournalIdentityHasNoAbsolutePath, redactAbsolutePaths, } from '../../../providers/identity/absolute-paths.js';
 import { CLAUDE_OBSERVATION_SOURCE } from '../../../providers/identity/claude-runtime-observation.js';
 import { CODEX_OBSERVATION_SOURCE } from '../../../providers/identity/codex-runtime-observation.js';
@@ -127,12 +128,34 @@ export function parseDispatchRecordInput(value) {
         throw new Error('Dispatch record input must be a JSON object.');
     }
     const keys = Object.keys(value).sort();
-    if (keys.length !== 2 || keys[0] !== 'event' || keys[1] !== 'record') {
-        throw new Error('Dispatch record input accepts only record and event.');
+    const isGenericInput = keys.length === 2 && keys[0] === 'event' && keys[1] === 'record';
+    const isManagedClaudeInput = keys.length === 3 &&
+        keys[0] === 'claudeLaunch' &&
+        keys[1] === 'event' &&
+        keys[2] === 'recordBase';
+    if (!isGenericInput && !isManagedClaudeInput) {
+        throw new Error('Dispatch record input accepts record and event, or claudeLaunch, event, and recordBase.');
     }
     if (!isRecord(value.event)) {
         throw new Error('Dispatch record event must be a JSON object.');
     }
+    const record = isManagedClaudeInput
+        ? buildClaudeDispatchRecord({
+            envelope: acceptClaudeLaunchEnvelope({
+                resolution: isRecord(value.claudeLaunch)
+                    ? value.claudeLaunch.resolution
+                    : undefined,
+                definition: isRecord(value.claudeLaunch)
+                    ? String(value.claudeLaunch.definition ?? '')
+                    : '',
+                launch: isRecord(value.claudeLaunch)
+                    ? value.claudeLaunch.payload
+                    : undefined,
+            }),
+            recordBase: value.recordBase,
+        })
+        : parseGenericDispatchRecord(value.record);
+    const boundedInput = { record, event: value.event };
     // Raw provider metadata is the one input a caller may legitimately supply
     // unmodified, so it is projected through the owning parser's allowlist and
     // asserted as the projection. Everything else is asserted exactly as given.
@@ -143,15 +166,14 @@ export function parseDispatchRecordInput(value) {
         ? value.event.metadata
         : null;
     assertNoSensitiveDispatchContent(rawMetadata === null
-        ? value
+        ? boundedInput
         : {
-            ...value,
+            ...boundedInput,
             event: {
                 ...value.event,
                 metadata: { ...rawMetadata, entries: null },
             },
         });
-    const record = parseGenericDispatchRecord(value.record);
     if (rawMetadata !== null) {
         if ('observation' in value.event) {
             throw new Error('A runtime observation event carries either observation or metadata, not both.');

package/dist/commands/project/dispatch-ceiling/index.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"index.d.ts","sourceRoot":"","sources":["../../../../src/commands/project/dispatch-ceiling/index.ts"],"names":[],"mappings":"AAGA,OAAO,EAEL,KAAK,cAAc,EACnB,KAAK,aAAa,EACnB,MAAM,sBAAsB,CAAC;AAkB9B,OAAO,EAUL,KAAK,uBAAuB,EAY7B,MAAM,oBAAoB,CAAC;AAE5B,OAAO,EAEL,KAAK,cAAc,EAEpB,MAAM,iBAAiB,CAAC;AA6BzB,OAAO,EAAE,OAAO,EAAU,MAAM,WAAW,CAAC;AAiC5C,UAAU,2BAA2B;IACnC,mBAAmB,EAAE,CAAC,OAAO,EAAE,aAAa,KAAK,cAAc,CAAC;IAChE,kBAAkB,EAAE,CAAC,GAAG,EAAE,MAAM,KAAK,OAAO,CAAC,MAAM,CAAC,CAAC;IACrD,sBAAsB,EAAE,CACtB,QAAQ,EAAE,MAAM,EAChB,aAAa,EAAE,MAAM,EACrB,GAAG,EAAE,MAAM,CAAC,UAAU,KACnB,OAAO,CAAC,cAAc,CAAC,CAAC;IAC7B,oBAAoB,EAAE,CAAC,QAAQ,EAAE,MAAM,KAAK,OAAO,CAAC,uBAAuB,CAAC,CAAC;IAC7E,QAAQ,EAAE,CAAC,IAAI,EAAE,MAAM,KAAK,OAAO,CAAC,MAAM,CAAC,CAAC;IAC5C,UAAU,EAAE,CAAC,IAAI,EAAE,MAAM,KAAK,OAAO,CAAC,OAAO,CAAC,CAAC;IAC/C,UAAU,EAAE,MAAM,CAAC,UAAU,CAAC;CAC/B;AAkyFD,wBAAgB,mCAAmC,CACjD,SAAS,GAAE,OAAO,CAAC,2BAA2B,CAAM,GACnD,OAAO,CA0GT"}
+{"version":3,"file":"index.d.ts","sourceRoot":"","sources":["../../../../src/commands/project/dispatch-ceiling/index.ts"],"names":[],"mappings":"AAGA,OAAO,EAEL,KAAK,cAAc,EACnB,KAAK,aAAa,EACnB,MAAM,sBAAsB,CAAC;AAkB9B,OAAO,EAUL,KAAK,uBAAuB,EAY7B,MAAM,oBAAoB,CAAC;AAE5B,OAAO,EAEL,KAAK,cAAc,EAEpB,MAAM,iBAAiB,CAAC;AAoCzB,OAAO,EAAE,OAAO,EAAU,MAAM,WAAW,CAAC;AAiC5C,UAAU,2BAA2B;IACnC,mBAAmB,EAAE,CAAC,OAAO,EAAE,aAAa,KAAK,cAAc,CAAC;IAChE,kBAAkB,EAAE,CAAC,GAAG,EAAE,MAAM,KAAK,OAAO,CAAC,MAAM,CAAC,CAAC;IACrD,sBAAsB,EAAE,CACtB,QAAQ,EAAE,MAAM,EAChB,aAAa,EAAE,MAAM,EACrB,GAAG,EAAE,MAAM,CAAC,UAAU,KACnB,OAAO,CAAC,cAAc,CAAC,CAAC;IAC7B,oBAAoB,EAAE,CAAC,QAAQ,EAAE,MAAM,KAAK,OAAO,CAAC,uBAAuB,CAAC,CAAC;IAC7E,QAAQ,EAAE,CAAC,IAAI,EAAE,MAAM,KAAK,OAAO,CAAC,MAAM,CAAC,CAAC;IAC5C,UAAU,EAAE,CAAC,IAAI,EAAE,MAAM,KAAK,OAAO,CAAC,OAAO,CAAC,CAAC;IAC/C,UAAU,EAAE,MAAM,CAAC,UAAU,CAAC;CAC/B;AAm4FD,wBAAgB,mCAAmC,CACjD,SAAS,GAAE,OAAO,CAAC,2BAA2B,CAAM,GACnD,OAAO,CA0GT"}

package/dist/commands/project/dispatch-ceiling/index.js

@@ -12,7 +12,9 @@ import { getOwnKey, setOwnKey } from '../../../config/own-keys.js';
 import { resolveEffectiveConfig, } from '../../../config/resolve.js';
 import { resolveProjectRoot } from '../../../fs/paths.js';
 import TOML from '@iarna/toml';
-import { getCeilingAdapter, isDirectDispatchRoleName, CLAUDE_TIER_ORDER, } from '../../../providers/ceiling/registry.js';
+import { getCeilingAdapter, isDirectDispatchRoleName, } from '../../../providers/ceiling/registry.js';
+import { claudeTargetRank, CLAUDE_EFFORT_ORDER, validateClaudeDispatchCapability, validateClaudeDispatchTarget, } from '../../../providers/claude/targets.js';
+import { SUPPORTED_CODEX_ROLE_TARGETS } from '../../../providers/codex/codec/shared.js';
 import { buildDispatchReport, formatDispatchReport, } from '../../../providers/identity/dispatch-report.js';
 import { classifyModelFamily, } from '../../../providers/identity/family.js';
 import { formatDispatchStamp, } from '../../../providers/identity/stamp.js';
@@ -21,6 +23,11 @@ import YAML from 'yaml';
 const CODEX_VALUES = [
     ...VALID_CODEX_DISPATCH_CEILINGS,
 ];
+// Catalogue order is low to high within each model; the scalar ceiling
+// remains separately capped by CODEX_VALUES.
+const CODEX_CANDIDATE_EFFORTS = [
+    ...new Set(SUPPORTED_CODEX_ROLE_TARGETS.map((target) => target.effort)),
+];
 const CLAUDE_VALUES = [
     ...VALID_CLAUDE_DISPATCH_CEILINGS,
 ];
@@ -124,6 +131,9 @@ function validManagedPolicyList() {
     return managedDispatchPolicyValueList(', ');
 }
 function validProviderValueList(provider) {
+    if (provider === 'codex') {
+        return CODEX_VALUES.join(', ');
+    }
     return providerValueOrder(provider)?.join(', ') ?? 'none';
 }
 function compiledPolicyValueForProvider(provider, compiled) {
@@ -259,6 +269,39 @@ function routeTargetFromObject(provider, target, routeIndex, routeLength) {
         routeLength,
     };
 }
+function resolveClaudeTargetCapability(target, env) {
+    if (!target ||
+        target.harness !== 'claude' ||
+        !target.model ||
+        !target.effort) {
+        return target;
+    }
+    const validation = validateClaudeDispatchCapability({
+        model: target.model,
+        effort: target.effort,
+        ...(target.capabilityEvidence
+            ? { capabilityEvidence: target.capabilityEvidence }
+            : {}),
+    }, env);
+    if (!validation.valid || !validation.capabilityEvidence) {
+        throw new Error(validation.reason ??
+            'Claude effort target has no established model capability.');
+    }
+    return { ...target, capabilityEvidence: validation.capabilityEvidence };
+}
+function resolveClaudePolicyCapabilities(policy, env) {
+    if (!policy)
+        return policy;
+    return {
+        ...policy,
+        target: resolveClaudeTargetCapability(policy.target, env) ?? null,
+        ...(policy.ceilingTarget !== undefined
+            ? {
+                ceilingTarget: resolveClaudeTargetCapability(policy.ceilingTarget, env) ?? null,
+            }
+            : {}),
+    };
+}
 function resolveRouteMatrixCell(provider, route, escalationLevel, cellSource) {
     const routeIndex = Math.min(escalationLevel, route.length - 1);
     const entry = route[routeIndex];
@@ -319,6 +362,8 @@ function resolveProviderCellFromValue(provider, providerValue, tier, cellSource,
         };
     }
     if (isWorkflowDispatchCandidateLadder(cell)) {
+        if (provider === 'claude')
+            assertCandidateOrder(provider, tier, cell);
         const ceiling = cell.candidates.at(-1);
         if (ceiling === undefined) {
             return null;
@@ -417,7 +462,7 @@ function candidatePrimaryTarget(provider, tier, candidate) {
     if (targetAdapter.selectionAxis === 'model-effort' &&
         (!target.model ||
             !target.effort ||
-            !CODEX_VALUES.includes(target.effort))) {
+            !CODEX_CANDIDATE_EFFORTS.includes(target.effort))) {
         throw new Error(`Malformed ${provider} candidate ordering in ${tier}: Codex candidates require a model and supported effort.`);
     }
     if ((targetAdapter.selectionAxis === 'model' ||
@@ -425,6 +470,15 @@ function candidatePrimaryTarget(provider, tier, candidate) {
         !target.model) {
         throw new Error(`Malformed ${provider} candidate ordering in ${tier}: model-argument candidates require a model.`);
     }
+    if (target.harness === 'claude' && target.model) {
+        const validation = validateClaudeDispatchTarget({
+            model: target.model,
+            ...(target.effort ? { effort: target.effort } : {}),
+        });
+        if (!validation.valid) {
+            throw new Error(`Malformed ${provider} candidate ordering in ${tier}: ${validation.reason}`);
+        }
+    }
     return target;
 }
 function candidateTargetKey(target) {
@@ -450,17 +504,18 @@ function assertCandidateOrder(provider, tier, ladder) {
         }
         seen.add(key);
         if (target.harness === 'claude' && target.model) {
-            const rank = CLAUDE_TIER_ORDER.indexOf(target.model);
-            if (rank < 0) {
-                throw new Error(`Malformed ${provider} candidate ordering in ${tier}: unsupported Claude model ${JSON.stringify(target.model)}.`);
-            }
+            const [modelRank, effortRank] = claudeTargetRank({
+                model: target.model,
+                ...(target.effort ? { effort: target.effort } : {}),
+            });
+            const rank = modelRank * CLAUDE_EFFORT_ORDER.length + effortRank;
             if (previousClaudeRank !== null && rank < previousClaudeRank) {
                 throw new Error(`Malformed ${provider} candidate ordering in ${tier}: Claude candidates must be nondecreasing.`);
             }
             previousClaudeRank = rank;
         }
         if (target.harness === 'codex' && target.model && target.effort) {
-            const rank = CODEX_VALUES.indexOf(target.effort);
+            const rank = CODEX_CANDIDATE_EFFORTS.indexOf(target.effort);
             const previousRank = codexRanksByModel.get(target.model);
             if (previousRank !== undefined && rank < previousRank) {
                 throw new Error(`Malformed ${provider} candidate ordering in ${tier}: Codex efforts for ${target.model} must be nondecreasing.`);
@@ -478,16 +533,26 @@ function assertTierCeilingsNondecreasing(provider, previous, current, tier) {
     if (current.harness === 'claude' &&
         previous.model &&
         current.model &&
-        CLAUDE_TIER_ORDER.indexOf(current.model) <
-            CLAUDE_TIER_ORDER.indexOf(previous.model)) {
+        (() => {
+            const [previousModel, previousEffort] = claudeTargetRank({
+                model: previous.model,
+                ...(previous.effort ? { effort: previous.effort } : {}),
+            });
+            const [currentModel, currentEffort] = claudeTargetRank({
+                model: current.model,
+                ...(current.effort ? { effort: current.effort } : {}),
+            });
+            return (currentModel < previousModel ||
+                (currentModel === previousModel && currentEffort < previousEffort));
+        })()) {
         throw new Error(`Malformed ${provider} candidate ordering in ${tier}: named tier ceilings must be nondecreasing.`);
     }
     if (current.harness === 'codex' &&
         previous.model === current.model &&
         previous.effort &&
         current.effort &&
-        CODEX_VALUES.indexOf(current.effort) <
-            CODEX_VALUES.indexOf(previous.effort)) {
+        CODEX_CANDIDATE_EFFORTS.indexOf(current.effort) <
+            CODEX_CANDIDATE_EFFORTS.indexOf(previous.effort)) {
         throw new Error(`Malformed ${provider} candidate ordering in ${tier}: named tier ceilings must be nondecreasing.`);
     }
 }
@@ -495,10 +560,11 @@ function requestedCandidateMatches(provider, requested, target) {
     if (target.harness !== provider || target.model !== requested.model) {
         return false;
     }
-    return provider !== 'codex' || target.effort === requested.effort;
+    return ((provider !== 'codex' && provider !== 'claude') ||
+        target.effort === requested.effort);
 }
 function formatRequestedCandidate(provider, requested) {
-    return provider === 'codex'
+    return provider === 'codex' || requested.effort !== undefined
         ? `${requested.model}/${requested.effort}`
         : requested.model;
 }
@@ -867,7 +933,7 @@ function normalizeRole(value) {
 }
 function providerValueOrder(provider) {
     if (provider === 'codex') {
-        return CODEX_VALUES;
+        return CODEX_CANDIDATE_EFFORTS;
     }
     if (provider === 'claude') {
         return CLAUDE_VALUES;
@@ -887,7 +953,7 @@ function normalizePreferredValue(provider, value) {
         return null;
     }
     if (!isValidProviderValue(provider, normalized)) {
-        const validValues = order.join(', ');
+        const validValues = (provider === 'codex' ? CODEX_VALUES : order).join(', ');
         throw new Error(`Invalid preferred dispatch value "${normalized}" for ${provider}. Valid values: ${validValues}.`);
     }
     return normalized;
@@ -927,17 +993,23 @@ function normalizeRequestedCandidate(provider, role, options) {
         if (!effort) {
             throw new Error('--candidate-effort is required for an exact Codex candidate.');
         }
-        if (!CODEX_VALUES.includes(effort)) {
-            throw new Error(`Invalid Codex candidate effort "${effort}". Valid values: ${CODEX_VALUES.join(', ')}.`);
+        if (!CODEX_CANDIDATE_EFFORTS.includes(effort)) {
+            throw new Error(`Invalid Codex candidate effort "${effort}". Valid values: ${CODEX_CANDIDATE_EFFORTS.join(', ')}.`);
         }
         return { model, effort };
     }
-    if (effort) {
-        throw new Error(`--candidate-effort is only valid for Codex; ${provider} candidates use --candidate-model only.`);
+    if (provider === 'claude') {
+        const validation = validateClaudeDispatchTarget({
+            model,
+            ...(effort ? { effort } : {}),
+        });
+        if (!validation.valid) {
+            throw new Error(validation.reason);
+        }
+        return effort ? { model, effort } : { model };
     }
-    if (provider === 'claude' &&
-        !CLAUDE_VALUES.includes(model)) {
-        throw new Error(`Invalid Claude candidate model "${model}". Valid values: ${CLAUDE_VALUES.join(', ')}.`);
+    if (effort) {
+        throw new Error(`--candidate-effort is only valid for Codex or Claude; ${provider} candidates use --candidate-model only.`);
     }
     return { model };
 }
@@ -965,12 +1037,12 @@ function normalizeClassification(provider, role, options) {
         throw new Error(`Invalid task class "${taskClassValue}". Valid values: ${TASK_CLASSES.join(', ')}.`);
     }
     const taskEffort = options.taskEffort?.trim() ?? '';
-    if (hasTaskEffort && provider !== 'codex') {
-        throw new Error(`--task-effort is only valid for Codex; received ${provider}.`);
+    if (hasTaskEffort && provider !== 'codex' && provider !== 'claude') {
+        throw new Error(`--task-effort is only valid for Codex or Claude; received ${provider}.`);
     }
-    if (hasTaskEffort &&
-        !CODEX_VALUES.includes(taskEffort)) {
-        throw new Error(`Invalid Codex task effort "${taskEffort}". Valid values: ${CODEX_VALUES.join(', ')}.`);
+    const validTaskEfforts = provider === 'claude' ? CLAUDE_EFFORT_ORDER : CODEX_CANDIDATE_EFFORTS;
+    if (hasTaskEffort && !validTaskEfforts.includes(taskEffort)) {
+        throw new Error(`Invalid ${provider === 'claude' ? 'Claude' : 'Codex'} task effort "${taskEffort}". Valid values: ${validTaskEfforts.join(', ')}.`);
     }
     return {
         taskClass: hasTaskClass ? taskClassValue : null,
@@ -1153,7 +1225,7 @@ function codexEffortAxis(selection, dispatchArgs) {
  * the enforcement mode, mechanism, dispatch args, and verify-on-upgrade flag.
  * Mode is computed here at call time — it is never read from persisted state.
  */
-function buildProviderResolution(provider, policy, role, orchestratorTier, preferredValue) {
+function buildProviderResolution(provider, policy, role, orchestratorTier, preferredValue, env) {
     const adapter = getCeilingAdapter(provider);
     if (policy === null) {
         return {
@@ -1193,6 +1265,7 @@ function buildProviderResolution(provider, policy, role, orchestratorTier, prefe
         ? adapter.compileToDispatchArgs(dispatchValue, role, {
             orchestratorTier,
             target: selection.target,
+            env,
         })
         : null;
     let mode;
@@ -1208,12 +1281,16 @@ function buildProviderResolution(provider, policy, role, orchestratorTier, prefe
     return {
         value: policy.value,
         mode,
-        mechanism: adapter.mechanism,
+        mechanism: hasVariantDispatchArgs(dispatchArgs)
+            ? 'pinned-variant'
+            : adapter.mechanism,
         dispatchArgs,
         modelAxis: modelAxis(provider, selection, dispatchArgs),
         effortAxis: provider === 'codex'
             ? codexEffortAxis(selection, dispatchArgs)
-            : 'not-applicable',
+            : provider === 'claude' && selection.target?.effort
+                ? `selected:${selection.target.effort}`
+                : 'not-applicable',
         verifyOnDispatch: dispatchValue && !isCrossHarness
             ? adapter.verifyOnDispatch(dispatchValue, {
                 orchestratorTier,
@@ -1385,11 +1462,11 @@ async function resolveDispatchCeiling(context, dependencies, options) {
     const escalationLevel = normalizeEscalationLevel(options.escalationLevel);
     const policyResolution = readResolvedConfigCeiling(provider, resolvedConfig) ??
         (await resolveProjectStateCeiling(provider, projectPath, dependencies));
-    const resolvedValue = await resolveCeilingValue(provider, resolvedConfig, projectPath, dependencies, escalationLevel, role, preferredValue, requestedCandidate, ceilingTier);
+    const resolvedValue = resolveClaudePolicyCapabilities(await resolveCeilingValue(provider, resolvedConfig, projectPath, dependencies, escalationLevel, role, preferredValue, requestedCandidate, ceilingTier), dependencies.processEnv);
     for (const warning of resolvedValue?.warnings ?? []) {
         context.logger.warn(warning);
     }
-    const providerResolution = buildProviderResolution(provider, resolvedValue, role, orchestratorTier, preferredValue);
+    const providerResolution = buildProviderResolution(provider, resolvedValue, role, orchestratorTier, preferredValue, dependencies.processEnv);
     const providers = {
         [provider]: providerResolution,
     };
@@ -1714,7 +1791,7 @@ function writeHumanResolution(context, resolution) {
         }
     }
     else {
-        context.logger.info('Effort axis: not-applicable');
+        context.logger.info(`Effort axis: ${providerResolution?.effortAxis ?? 'not-applicable'}`);
         if (providerResolution?.selection.selectionMode === 'inherit-default') {
             context.logger.info('Note: OAT will not select a Claude model; Task dispatch inherits host/provider behavior.');
         }
@@ -1910,9 +1987,9 @@ export function createProjectDispatchCeilingCommand(overrides = {}) {
         .option('--preferred <value>', 'Legacy preferred implementer/fix value before applying the resolved policy')
         .addOption(new Option('--ceiling-tier <tier>', 'Invocation-only named maximum tier; never persists configuration or project state').choices([...VALID_DISPATCH_MATRIX_TIERS]))
         .option('--candidate-model <model>', 'Exact configured implementer candidate model beneath the named ceiling')
-        .option('--candidate-effort <effort>', 'Exact configured Codex candidate effort paired with --candidate-model')
+        .option('--candidate-effort <effort>', 'Exact configured Codex or Claude candidate effort paired with --candidate-model')
         .option('--task-class <class>', `Provider-neutral task classification provenance: ${TASK_CLASSES.join(', ')}`)
-        .option('--task-effort <effort>', 'Codex-only task effort classification provenance; does not select a candidate')
+        .option('--task-effort <effort>', 'Codex or Claude task effort classification provenance; does not select a candidate')
         .option('--escalation-level <level>', 'Ordered route entry to select; 0 is the floor and higher values advance through escalation targets')
         .option('--project-path <path>', 'Read project-state policy from an explicit project path')
         .option('--preflight', 'Treat unresolved non-interactive resolution as an implementation block')

package/dist/config/dispatch-matrix.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"dispatch-matrix.d.ts","sourceRoot":"","sources":["../../src/config/dispatch-matrix.ts"],"names":[],"mappings":"AAAA,MAAM,MAAM,0BAA0B,GAClC,SAAS,GACT,UAAU,GACV,MAAM,GACN,UAAU,CAAC;AAEf,MAAM,WAAW,2BAA2B;IAC1C,OAAO,CAAC,EAAE,MAAM,CAAC;IACjB,KAAK,CAAC,EAAE,MAAM,CAAC;IACf,MAAM,CAAC,EAAE,MAAM,CAAC;CACjB;AAED,MAAM,WAAW,6BAA6B;IAC5C,KAAK,EAAE,OAAO,CAAC;IACf,MAAM,CAAC,EAAE,MAAM,CAAC;CACjB;AAED,MAAM,MAAM,0BAA0B,GAAG,MAAM,GAAG,2BAA2B,CAAC;AAC9E,MAAM,MAAM,qBAAqB,GAAG,0BAA0B,EAAE,CAAC;AAEjE,MAAM,WAAW,6BAA6B;IAC5C,KAAK,EAAE,qBAAqB,CAAC;CAC9B;AAED,MAAM,MAAM,yBAAyB,GACjC,0BAA0B,GAC1B,6BAA6B,CAAC;AAElC,MAAM,WAAW,+BAA+B;IAC9C,UAAU,EAAE,yBAAyB,EAAE,CAAC;CACzC;AAED,MAAM,MAAM,gCAAgC,GAAG,MAAM,GAAG,qBAAqB,CAAC;AAC9E,MAAM,MAAM,0BAA0B,GAClC,+BAA+B,GAC/B,gCAAgC,CAAC;AACrC,MAAM,MAAM,6BAA6B,GACrC,MAAM,GACN,OAAO,CAAC,MAAM,CAAC,0BAA0B,EAAE,0BAA0B,CAAC,CAAC,CAAC;AAE5E,MAAM,MAAM,oCAAoC,GAC5C,oBAAoB,GACpB,gBAAgB,GAChB,qBAAqB,CAAC;AAE1B,MAAM,WAAW,gCAAgC;IAC/C,IAAI,EAAE,MAAM,CAAC;IACb,IAAI,EAAE,oCAAoC,CAAC;IAC3C,KAAK,EAAE,OAAO,CAAC;CAChB;AAED,MAAM,WAAW,wBAAwB;IACvC,SAAS,EAAE,MAAM,CAAC,MAAM,EAAE,6BAA6B,CAAC,CAAC;IACzD,MAAM,EAAE,gCAAgC,EAAE,CAAC;CAC5C;AAED,MAAM,MAAM,oBAAoB,GAC5B,cAAc,GACd,aAAa,GACb,aAAa,GACb,eAAe,CAAC;AAEpB,MAAM,WAAW,yBAAyB;IACxC,MAAM,EAAE,oBAAoB,CAAC;IAC7B,UAAU,EAAE,MAAM,CAAC;CACpB;AAED,MAAM,WAAW,qBAAqB;IACpC,QAAQ,EAAE,MAAM,CAAC;IACjB,IAAI,EAAE,0BAA0B,GAAG,IAAI,CAAC;IACxC,cAAc,EAAE,MAAM,GAAG,IAAI,CAAC;IAC9B,kBAAkB,EAAE,MAAM,GAAG,IAAI,CAAC;IAClC,KAAK,EAAE,MAAM,GAAG,IAAI,CAAC;IACrB,MAAM,EAAE,2BAA2B,GAAG,IAAI,CAAC;IAC3C,IAAI,EAAE,MAAM,CAAC;IACb,MAAM,EAAE,oBAAoB,CAAC;CAC9B;AAED,eAAO,MAAM,2BAA2B,EAAE,SAAS,0BAA0B,EAChC,CAAC;AAyM9C,wBAAgB,uBAAuB,CACrC,KAAK,EAAE,OAAO,EACd,OAAO,EAAE;IACP,UAAU,EAAE,MAAM,CAAC;IACnB,iBAAiB,EAAE,gBAAgB,GAAG,eAAe,CAAC;CACvD,GACA,wBAAwB,CA2E1B;AAED,wBAAgB,8BAA8B,CAC5C,QAAQ,EAAE,MAAM,EAChB,MAAM,EAAE,2BAA2B,GAClC,OAAO,CAET;AAED,wBAAgB,2BAA2B,CACzC,QAAQ,EAAE,MAAM,EAChB,MAAM,EAAE,2BAA2B,GAClC,6BAA6B,CAc/B;AAED,wBAAgB,+BAA+B,CAC7C,KAAK,EAAE,OAAO,GACb,KAAK,IAAI,6BAA6B,CAExC;AAED,wBAAgB,iCAAiC,CAC/C,KAAK,EAAE,OAAO,GACb,KAAK,IAAI,+BAA+B,CAE1C;AAED,wBAAgB,iCAAiC,CAC/C,IAAI,EAAE,0BAA0B,GAC/B,+BAA+B,CAQjC;AAwBD,wBAAgB,kBAAkB,CAChC,SAAS,EAAE,MAAM,CAAC,MAAM,EAAE,6BAA6B,CAAC,EACxD,OAAO,EAAE,yBAAyB,GACjC,qBAAqB,EAAE,CAyDzB"}
+{"version":3,"file":"dispatch-matrix.d.ts","sourceRoot":"","sources":["../../src/config/dispatch-matrix.ts"],"names":[],"mappings":"AAEA,MAAM,MAAM,0BAA0B,GAClC,SAAS,GACT,UAAU,GACV,MAAM,GACN,UAAU,CAAC;AAEf,MAAM,WAAW,2BAA2B;IAC1C,OAAO,CAAC,EAAE,MAAM,CAAC;IACjB,KAAK,CAAC,EAAE,MAAM,CAAC;IACf,MAAM,CAAC,EAAE,MAAM,CAAC;CACjB;AAED,MAAM,WAAW,6BAA6B;IAC5C,KAAK,EAAE,OAAO,CAAC;IACf,MAAM,CAAC,EAAE,MAAM,CAAC;CACjB;AAED,MAAM,MAAM,0BAA0B,GAAG,MAAM,GAAG,2BAA2B,CAAC;AAC9E,MAAM,MAAM,qBAAqB,GAAG,0BAA0B,EAAE,CAAC;AAEjE,MAAM,WAAW,6BAA6B;IAC5C,KAAK,EAAE,qBAAqB,CAAC;CAC9B;AAED,MAAM,MAAM,yBAAyB,GACjC,0BAA0B,GAC1B,6BAA6B,CAAC;AAElC,MAAM,WAAW,+BAA+B;IAC9C,UAAU,EAAE,yBAAyB,EAAE,CAAC;CACzC;AAED,MAAM,MAAM,gCAAgC,GAAG,MAAM,GAAG,qBAAqB,CAAC;AAC9E,MAAM,MAAM,0BAA0B,GAClC,+BAA+B,GAC/B,gCAAgC,CAAC;AACrC,MAAM,MAAM,6BAA6B,GACrC,MAAM,GACN,OAAO,CAAC,MAAM,CAAC,0BAA0B,EAAE,0BAA0B,CAAC,CAAC,CAAC;AAE5E,MAAM,MAAM,oCAAoC,GAC5C,oBAAoB,GACpB,gBAAgB,GAChB,qBAAqB,CAAC;AAE1B,MAAM,WAAW,gCAAgC;IAC/C,IAAI,EAAE,MAAM,CAAC;IACb,IAAI,EAAE,oCAAoC,CAAC;IAC3C,KAAK,EAAE,OAAO,CAAC;CAChB;AAED,MAAM,WAAW,wBAAwB;IACvC,SAAS,EAAE,MAAM,CAAC,MAAM,EAAE,6BAA6B,CAAC,CAAC;IACzD,MAAM,EAAE,gCAAgC,EAAE,CAAC;CAC5C;AAED,MAAM,MAAM,oBAAoB,GAC5B,cAAc,GACd,aAAa,GACb,aAAa,GACb,eAAe,CAAC;AAEpB,MAAM,WAAW,yBAAyB;IACxC,MAAM,EAAE,oBAAoB,CAAC;IAC7B,UAAU,EAAE,MAAM,CAAC;CACpB;AAED,MAAM,WAAW,qBAAqB;IACpC,QAAQ,EAAE,MAAM,CAAC;IACjB,IAAI,EAAE,0BAA0B,GAAG,IAAI,CAAC;IACxC,cAAc,EAAE,MAAM,GAAG,IAAI,CAAC;IAC9B,kBAAkB,EAAE,MAAM,GAAG,IAAI,CAAC;IAClC,KAAK,EAAE,MAAM,GAAG,IAAI,CAAC;IACrB,MAAM,EAAE,2BAA2B,GAAG,IAAI,CAAC;IAC3C,IAAI,EAAE,MAAM,CAAC;IACb,MAAM,EAAE,oBAAoB,CAAC;CAC9B;AAED,eAAO,MAAM,2BAA2B,EAAE,SAAS,0BAA0B,EAChC,CAAC;AAyM9C,wBAAgB,uBAAuB,CACrC,KAAK,EAAE,OAAO,EACd,OAAO,EAAE;IACP,UAAU,EAAE,MAAM,CAAC;IACnB,iBAAiB,EAAE,gBAAgB,GAAG,eAAe,CAAC;CACvD,GACA,wBAAwB,CA2E1B;AAED,wBAAgB,8BAA8B,CAC5C,QAAQ,EAAE,MAAM,EAChB,MAAM,EAAE,2BAA2B,GAClC,OAAO,CAET;AAED,wBAAgB,2BAA2B,CACzC,QAAQ,EAAE,MAAM,EAChB,MAAM,EAAE,2BAA2B,GAClC,6BAA6B,CA2B/B;AAED,wBAAgB,+BAA+B,CAC7C,KAAK,EAAE,OAAO,GACb,KAAK,IAAI,6BAA6B,CAExC;AAED,wBAAgB,iCAAiC,CAC/C,KAAK,EAAE,OAAO,GACb,KAAK,IAAI,+BAA+B,CAE1C;AAED,wBAAgB,iCAAiC,CAC/C,IAAI,EAAE,0BAA0B,GAC/B,+BAA+B,CAQjC;AAwBD,wBAAgB,kBAAkB,CAChC,SAAS,EAAE,MAAM,CAAC,MAAM,EAAE,6BAA6B,CAAC,EACxD,OAAO,EAAE,yBAAyB,GACjC,qBAAqB,EAAE,CAyDzB"}