axstack 0.57.0 → 0.58.0

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "axstack",
-  "version": "0.57.0",
+  "version": "0.58.0",
   "description": "Axstack installer and setup CLI: installs owned chat skills and role data, configures supported harness settings, and checks T3 Code capabilities.",
   "keywords": [
     "claude-code",

package/profiles/presets/claude-only.json

@@ -297,7 +297,7 @@
       "model": null,
       "modeId": "bypassPermissions",
       "thinkingOptionId": "high",
-      "notes": "Intentional single-provider absence in the claude-only preset; Grok candidate unavailable. Arena holds without substitution."
+      "notes": "Record the intentional single-provider absence of optional Grok in claude-only. Never substitute another model. Required candidate and judge absences still hold their applicable question or round."
     }
   ]
 }

package/profiles/presets/codex-only.json

@@ -296,7 +296,7 @@
       "model": null,
       "modeId": "full-access",
       "thinkingOptionId": "high",
-      "notes": "Intentional single-provider absence in the codex-only preset; Grok candidate unavailable. Arena holds without substitution."
+      "notes": "Record the intentional single-provider absence of optional Grok in codex-only. Never substitute another model. Required candidate and judge absences still hold their applicable question or round."
     }
   ]
 }

package/profiles/presets/mixed.json

@@ -299,7 +299,7 @@
       "model": null,
       "modeId": "full-access",
       "thinkingOptionId": "high",
-      "notes": "Independent Grok Rung 2 design candidate. Use the T3 Grok provider and resolve the model from saved capabilities. Receives the same brief without cross-reading; returns design, rationale, and rejected alternatives."
+      "notes": "Independent optional Grok design candidate for every light arena. Use the T3 Grok provider and resolve the model from saved capabilities. Receives the same brief without cross-reading; returns design, rationale, and rejected alternatives."
     }
   ]
 }

package/skills/axstack/references/contracts.md

@@ -98,6 +98,11 @@ to the user with no reconciliation and no Fable.
 
 ## Advised holds
 
+For an eligible local nonrepo report-only consultation, follow the explicit
+advice workflow in [Local artifacts](local-artifacts.md). Bind its actual parent
+phase and trigger in the brief; this changes no parent prerequisite, the
+never-advised exclusions below, configured seats or escalation conditions.
+
 Advise when a hold reaches the user and the reply picks between two or more
 paths an agent can then carry out.
 The criterion governs and the examples are illustrative.

package/skills/axstack/references/fleet-dispatch.md

@@ -7,7 +7,7 @@ records and forge actions. Helpers validate explicit saved inputs only: no
 network, clock, dispatch, retry, merge, retained state or aggregate permission grant.
 Synthetic evidence can pass binding checks; helpers cannot authenticate it.
 
-For explicit local non-repository Research/Explain/Publish work, apply
+For supported local non-repository report-only work, apply
 [Local artifacts](local-artifacts.md): its discriminated scope, native read/list
 null-identity proof and owned folders replace only the repository/checkout terms.
 All native execution authority and independent exact-byte review remain required.
@@ -130,6 +130,8 @@ attempt (key, exact title, clientRequestId, origin environmentId/threadId/taskId
 candidate/base SHAs for repositories, artifact SHA-256/null base for local artifacts),
 requested target/runtimeMode/isolation and native submission. Artifact ownership
 and byte readbacks follow [Local artifacts](local-artifacts.md#receipts).
+That reference also defines the local-only T3 0.1.5 child-title projection;
+the attempt always retains the full exact requested title.
 The driver's independent fleetAttempt is the expected run-record identity;
 never copy it from an arriving receipt to make a comparison pass.
 

package/skills/axstack/references/fleet-route-schema.md

@@ -116,6 +116,10 @@ for serial emulator use, desktop safety and declared thermal prechecks.
 legacy omitted scope retains that repository requirement. The explicit
 `{kind:"local-artifact", ...}` alternative requires repositoryId:null and the
 workflow, role, native project and ownership fields in [Local artifacts](local-artifacts.md).
+The local-artifact matrix separates direct and inherited roles for research, explain,
+publish, brainstorm, align, improve, audit and conditional advice.
+Brainstorm candidate/judge roles never leak into source/article authority.
+Applicable rung/round duties remain report-only procedural checks, not route authority.
 Artifact destination project also carries workspaceRoot; only the bound local
 environment/project/workspace is eligible. Null alone never selects a route.
 

package/skills/axstack/references/lifecycle.md

@@ -29,10 +29,12 @@ Prefer parallel independent bounded work; no redundant workers. Fanout is depend
 capacity-driven within host/spending limits. [PR shape](pr-shape.md) covers
 theme/size; queue dependencies under [PR chains](pr-chain.md).
 
-For explicit non-repository Research/Explain/Publish artifacts, use
+For supported non-repository report-only workflows, use
 [Local artifacts](local-artifacts.md) for owned shared delegates and SHA-256
 receipts. The reviewer checkout row above applies to code candidates; article
 review retains independent source/fact and exact-byte UI/design checks.
+Brainstorm candidates and applicable judges retain independent report-only
+arena duties and conditional rounds; this grants no engineering authority.
 Null identity does not waive engineering isolation or authorize remote artifacts.
 
 ## Ownership
@@ -182,6 +184,10 @@ Build no custom scheduler, state engine or legacy fallback. See
 Audit measurement is enabled by default for every substantive run; dispatch
 follows [Close-out](#close-out) or a material-deviation/repeated-repair
 checkpoint. An `axstack-audit` run is excluded; it launches no children.
+For supported nonrepo report runs, keep the originating workflow and inherited
+auditor roles in [Local artifacts](local-artifacts.md); standalone audit uses
+workflow audit and never audits itself. Conditional advice uses that matrix
+and the standing advised-hold/escalation rules, without waiving parent prerequisites.
 Load [axstack-audit](../../axstack-audit/SKILL.md). Accepted proposals
 require a regression scenario and unchanged holdout checks; they change
 nothing without tested independent review.

package/skills/axstack/references/local-artifacts.md

@@ -1,7 +1,8 @@
 # Local non-repository artifacts
 
-Research, Explain and Publish may use an explicitly bound non-repository
-artifact scope. Null repository identity alone never authorizes a route.
+The report-only modes in the matrix below may use an explicitly bound
+non-repository artifact scope. Null repository identity alone never authorizes
+a route.
 Engineering authors and code reviewers still require a repository and native
 Git worktree isolation. Remote artifacts and remote author/coordinator work
 remain outside this scope.
@@ -9,8 +10,9 @@ remain outside this scope.
 ## Scope and ownership
 
 Use `request.scope` below with `repositoryId:null` and `role:reader`.
-Here reader means no product-code authority; an explainer may write only its
-owned article artifacts. `task` remains a load preference, not role authority.
+Here reader means no product-code authority. Writers may write only their owned
+artifacts or reports under the selected phase contract.
+`task` remains a load preference, not role authority.
 The scope is evidence for the helpers and brief, never a native T3 argument.
 
 ```json
@@ -27,8 +29,8 @@ The scope is evidence for the helpers and brief, never a native T3 argument.
 }
 ```
 
-`workflow` is research, explain or publish. All three permit the configured
-research requirements/code/code-sol/web/web-sol/X and explore-execution/
+The research, explain and publish workflows permit the configured research
+requirements/code/code-sol/web/web-sol/X and explore-execution/
 explore-execution-sol source roles plus diligence. Explain and Publish also
 permit explainer, explainer-review and the configured UI verifier roles.
 Use their full `axstack-` role IDs; this list does not make an absent seat
@@ -36,6 +38,91 @@ available or change provider/model/effort/options/runtime authority.
 The selected role ID must equal scope.roleId in dispatch-plan.
 No engineering author, code reviewer, owner or arbitrary role can use this scope.
 
+Brainstorm permits only candidate seats Astra (`axstack-advisor-astra`),
+Opus (`axstack-advisor-opus`), Grok (`axstack-arena-candidate-grok`) and judge
+seats Opus (`axstack-arena-judge-opus`), Fable (`axstack-escalation-fable`), Astra
+(`axstack-arena-judge-astra`). All outputs are report-only; never grant
+implementation or spec approval. These seats are excluded from Research,
+Explain and Publish artifact scopes; source/article roles are excluded from
+Brainstorm. These are direct roles; inherited audit is listed below.
+This is arena authority, never arbitrary adviser authority.
+
+Candidates must remain independent with the same bounded brief and no
+cross-reading; judges never author candidates or cross-read other judges.
+Rung 1 has no judge round; Rung 2 uses round 1, with round 2 only on the
+existing disagreement or caller's user-rejection trigger in
+[the arena procedure](../../axstack-brainstorm/references/arena.md).
+The driver verifies rung, round, independence and configured seat availability
+before dispatch; helpers validate role scope, not those procedural facts.
+
+Use the selected preset's exact provider/model/effort/full-access bindings.
+Configured optional Grok follows the arena's fenced absence rule; an uncertain
+dispatch requires reconciliation. Fable is a required round-2 judge when that
+round applies, not an optional candidate. A missing required model or an
+intentional null-model absence holds the applicable round without substitution.
+An allowlisted role never makes an unavailable seat executable.
+
+## Workflow and inherited-role matrix
+
+Substantive parent runs in each supported workflow retain `axstack-auditor` and
+`axstack-auditor-sol` under the lifecycle hook. Their passes must remain
+independent on the same bounded evidence, with configured optional-seat and
+mixed-provider rules unchanged. Audit never audits itself; its auditor workers
+launch no children. The unlaunchable-base UNKNOWN archive exception remains
+a degraded receipt, not evidence of a completed audit.
+
+| Workflow | Direct roles | Inherited audit | Boundary |
+| --- | --- | --- | --- |
+| research | Source roles listed above | Both auditors | Cited, bounded research; no engineering |
+| explain, publish | Source and article roles listed above | Both auditors | Exact-byte review; publication still needs authority |
+| brainstorm | Six candidate/judge seats listed above | Both auditors | Independent report-only arena and applicable rounds |
+| align | `axstack-advisor-astra`, `axstack-advisor-opus` | Both auditors | Pre-repository scope advice; stop before execution |
+| improve | `axstack-research-code`, `axstack-research-code-sol`, `axstack-ui-verifier`, `axstack-ui-verifier-sonnet` | Both auditors | Named nonversioned code/UI discovery only; second UI seat only under its dual-run trigger |
+| audit | `axstack-auditor`, `axstack-auditor-sol` | None; terminal audit | Standalone audit of explicit nonrepo evidence; only assigned `audit.md` is writable |
+| advice | `axstack-advisor-astra`, `axstack-advisor-opus`, `axstack-escalation-fable` | None; audit retains the supported parent's workflow | Only eligible existing shared advice; Fable retains its trigger |
+
+The advice workflow permits only existing advised-hold, high-stakes or bounded
+escalation consultations under [Standing contracts](contracts.md). It never
+waives the parent phase's repository, store or tracker prerequisites. Record
+the actual parent phase, qualifying trigger, evidence and question in the brief.
+Ordinary advised holds use Astra/Opus; high-stakes use Astra/Fable; other Fable
+uses require the existing bounded escalation trigger and a fresh session.
+Never relabel arbitrary work as advice. Required seats and missing models hold
+without substitution. The helper checks the role matrix; the driver must verify
+these conditional duties, as with arena rounds.
+
+Nested phase calls use their own explicit scope: Align research uses research,
+unresolved design uses brainstorm, and Improve visuals use explain. This is not
+a workflow alias for the caller or a waiver of its prerequisites.
+
+Improve already accepts nonversioned files, screenshots and exported snippets.
+Its previous artifact routing was ambiguous, not a proven runtime failure.
+The improve mapping now covers that existing report-only discovery: copy owned
+source snapshots, bind stable source identity and SHA-256, and name missing
+history. UI discovery retains exact build/URL, source identity, real Playwright
+capture and applicable dual-run requirements. A supplied screenshot alone is
+not proof of a live render. Repository readiness, whole-repo Scout, executable
+probes and implementation do not acquire artifact authority.
+
+| Skill or mode | Nonrepo boundary |
+| --- | --- |
+| Anatomy | Uses Explain's workflow and exact-byte review; no new workflow |
+| Perf | Router only; Improve discovery may use improve, Debug/Implement still require repository isolation |
+| Spec | Unsupported as a standalone artifact workflow; chosen authoritative store, repository counterpart and approval remain required |
+| Tickets | Unsupported; approved engineering scope, repository/store and tracker access remain required |
+| Scout | Unsupported; whole-repository pinned-tree evidence is required |
+| Debug, Implement | Unsupported; native repository/worktree probes and writers remain required |
+| Review | Article/source review uses explain/publish; PR/code/codebase review retains repository and checkout requirements |
+| Verify | Unsupported; repository verification-skill authoring routes through Implement |
+| Watch | Unsupported; PR/repository ownership and forge rules remain required |
+| Correct, Cleanup, VOA | Inline only; no worker workflow added; existing identity, ownership and read-only/status limits remain |
+
+This matrix plus the supported rows covers the twenty skills inspected for this
+repair. It describes contract/validator support, not completed native execution.
+No live provider, browser, tracker, publication or installed-host support follows
+from static checks or fixture receipts. Missing prerequisites remain unresolved;
+record the exact gap rather than fabricate a repository, alias or approval.
+
 The driver binds the native parent project and its real workspaceRoot. Require
 explicit null repositoryIdentity in both project-read and complete project-list
 reads; omitted, conflicting or incomplete observations hold. Native environment,
@@ -77,6 +164,12 @@ Publication remains the explicitly authorized driver's action.
 ## Receipts
 
 Retain the normal native task/thread/run, pending-question and settings checks.
+For audits, the driver supplies an owned read-only input snapshot or evidence
+manifest under artifactRoot. Bind its SHA-256 as the candidate; the auditor
+writes only evidenceRoot/audit.md and marker report points there. Missing Git
+history is a named limitation, never a fabricated revision. Paired auditors get
+independent owned snapshots and never cross-read reports.
+
 For artifact receipts, `attempt.candidate` and marker `head` are the exact
 64-hex artifact SHA-256; `attempt.base` is null. They are not Git revisions.
 Bind `artifactScope` to the requested scope and report to evidenceRoot.
@@ -97,4 +190,13 @@ run and its bytes equal the destination final message. An explicit conflicting
 terminal summary, stale run, other version or remote result does not use this
 fallback. Preserve the raw native fields; do not rewrite them. Never synthesize an effective field from requested settings.
 The driver checks actual bytes and readbacks; helpers validate saved data only.
+For local artifact receipts on observed T3 0.1.5 only, a trimmed requested title
+longer than 72 UTF-16 code units may appear on the child as its first 69 units
+plus three ASCII periods (`...`). Compare that exact native projection, never
+an arbitrary prefix or Unicode ellipsis. Keep the full requested title in the
+independent attempt; do not rewrite either native or expected evidence. Task,
+child, run, parent, marker and artifact bindings still apply: clipped titles are
+not unique identity. Other versions, remote and engineering receipts retain
+exact title equality. Saved replay proves compatibility, not a fresh dispatch.
+
 Keep artifact receipts separate from Git candidate publication/PR/merge receipts.

package/skills/axstack/references/role-roster.md

@@ -6,7 +6,7 @@ native destination. Remote reviewers require T3 worktree isolation; no shell chi
 checkout. Host names confer no role or permission. Remote authors/coordinator stay
 disabled; local isolated authors remain supported even overloaded.
 
-Explicit non-repository Research/Explain/Publish uses the bounded role allowlist
+Supported non-repository report-only workflows use the bounded role matrix
 and local shared-artifact isolation in [Local artifacts](local-artifacts.md).
 This changes no configured seat, source independence or exact-byte review duty.
 Engineering authors and code reviewers cannot acquire this artifact authority.

package/skills/axstack/references/t3-runtime.md

@@ -204,10 +204,10 @@ a candidate follows [UI verification](ui-verification.md).
 
 The driver must be the sole run-record writer and enforce one writer per
 candidate; it never writes tracked candidate source or tests or repairs an author's source.
-For explicit non-repository Research/Explain/Publish only,
+For supported non-repository report-only workflows only,
 [Local artifacts](local-artifacts.md) supplies the artifact branch of this table:
-source readers, diligence, explainers and article/UI reviewers use local shared
-native delegates with separate owned folders and exact SHA-256 snapshots.
+Only the matrix's direct and inherited roles use local shared native delegates
+with separate owned folders and exact SHA-256 snapshots.
 Its role allowlist excludes engineering authors and code reviewers; their native
 Git worktree rules above remain mandatory. Null repository alone grants nothing.
 

package/skills/axstack/references/ui-verification.md

@@ -24,7 +24,7 @@ All other browser checks outside the two named exceptions remain delegated to co
 
 Read the [T3 runtime boundary](t3-runtime.md) before dispatch; for candidate
 checks use its driver-made disposable detached checkout at the pinned candidate SHA.
-For explicit non-repository article checks, use the local shared delegate and
+For explicit non-repository article or Improve discovery checks, use the local shared delegate and
 owned SHA-256 snapshot in [Local artifacts](local-artifacts.md); no Git candidate
 is fabricated. Engineering candidate checks retain the checkout requirement.
 Use the configured verifier browser and honor the user browser-tool policy.

package/skills/axstack/scripts/fleet-receipt.js

@@ -41,7 +41,12 @@ export function inspectFleetReceipt(receipt, expected) {
     && r.thread?.environmentId === s.environmentId && thread?.threadId === child(status),
   { submitted: submitted?.remoteChild ?? submitted?.childThreadId,
     status: status?.remoteChild ?? status?.childThreadId, thread: r.thread?.environmentId });
-  term('receipt-title', text(a?.title) && thread?.title === a.title, thread?.title);
+  // T3 0.1.5 persists subagentThreadTitle's exact projection on local children.
+  // Full attempt/task/run/marker identity remains independently mandatory.
+  const projectedTitle = artifact && !remote && s.versions?.t3 === '0.1.5'
+    && text(a?.title) && a.title === a.title.trim() && a.title.length > 72
+    && thread?.title === a.title.slice(0, 69) + '...';
+  term('receipt-title', text(a?.title) && (thread?.title === a.title || projectedTitle), thread?.title);
   term('receipt-project', thread?.projectId === s.projectId, thread?.projectId);
   for (const [name, tool, environmentId, args] of [
     ['status', 'task_status', a?.origin?.environmentId, { taskId: a?.origin?.taskId }],

package/skills/axstack/scripts/fleet-scope.js

@@ -3,6 +3,20 @@ const sourceRoles = ['axstack-research-requirements', 'axstack-research-code', '
   'axstack-explore-execution', 'axstack-explore-execution-sol',
   'axstack-research-web', 'axstack-research-web-sol', 'axstack-research-x', 'axstack-diligence'];
 const articleRoles = ['axstack-explainer', 'axstack-explainer-review', 'axstack-ui-verifier', 'axstack-ui-verifier-sonnet'];
+const brainstormRoles = ['axstack-advisor-astra', 'axstack-advisor-opus', 'axstack-arena-candidate-grok',
+  'axstack-arena-judge-opus', 'axstack-escalation-fable', 'axstack-arena-judge-astra'];
+const auditRoles = ['axstack-auditor', 'axstack-auditor-sol'];
+const workflowRoles = {
+  research: [...sourceRoles, ...auditRoles],
+  explain: [...sourceRoles, ...articleRoles, ...auditRoles],
+  publish: [...sourceRoles, ...articleRoles, ...auditRoles],
+  brainstorm: [...brainstormRoles, ...auditRoles],
+  align: ['axstack-advisor-astra', 'axstack-advisor-opus', ...auditRoles],
+  improve: ['axstack-research-code', 'axstack-research-code-sol', 'axstack-ui-verifier',
+    'axstack-ui-verifier-sonnet', ...auditRoles],
+  audit: auditRoles,
+  advice: ['axstack-advisor-astra', 'axstack-advisor-opus', 'axstack-escalation-fable'],
+};
 export const isArtifact = (request) => request?.scope?.kind === 'local-artifact';
 const path = (value) => typeof value === 'string' && value.startsWith('/') && value !== '/'
   && !/[\s\x00-\x1f\x7f]/.test(value) && !value.slice(1).split('/').some((v) => ['', '.', '..'].includes(v));
@@ -14,8 +28,7 @@ export function validScope(request) {
   if (s === undefined) return typeof request.repositoryId === 'string' && request.repositoryId.trim().length > 0;
   if (s?.kind === 'repository') return typeof request.repositoryId === 'string' && request.repositoryId.trim().length > 0;
   if (!isArtifact(request) || request.repositoryId !== null || request.role !== 'reader'
-    || !['research', 'explain', 'publish'].includes(s.workflow)
-    || ![...sourceRoles, ...(s.workflow === 'research' ? [] : articleRoles)].includes(s.roleId)
+    || typeof s.workflow !== 'string' || !Object.hasOwn(workflowRoles, s.workflow) || !workflowRoles[s.workflow].includes(s.roleId)
     || ![s.environmentId, s.projectId].every((v) => typeof v === 'string' && v.trim().length > 0)
     || !inside(s.artifactRoot, s.workspaceRoot) || !inside(s.evidenceRoot, s.workspaceRoot)
     || overlaps(s.artifactRoot, s.evidenceRoot) || !Array.isArray(s.reservedRoots)

package/skills/axstack-align/SKILL.md

@@ -8,6 +8,11 @@ description: When exploring or planning engineering work, use axstack-align to s
 For authorized delivery runs, follow [Autopilot](../axstack/references/autopilot.md)
 for phase continuation and holds.
 
+Before repository binding, report-only Align advice uses workflow align in
+[Local artifacts](../axstack/references/local-artifacts.md). Research and
+Brainstorm retain their own scopes; inherited audit and eligible shared advice
+follow the same matrix. This grants no Spec/store, tracker or execution authority.
+
 On driver entry, sweep under [Workspace hygiene](../axstack/references/workspace-hygiene.md); dispatched workers do not sweep.
 For every dispatch brief, name its private `<run dir>/evidence/<dispatch>/` folder.
 

package/skills/axstack-audit/SKILL.md

@@ -5,6 +5,12 @@ description: When a substantive run needs evidence review, use axstack-audit to
 
 # Audit
 
+For explicit nonrepo evidence, follow
+[Local artifacts](../axstack/references/local-artifacts.md): standalone audit uses
+workflow audit; inherited audits retain the supported parent's workflow. The
+driver binds owned input snapshots and SHA-256; only the assigned audit.md is
+writable. Missing Git history stays a limitation, never a fabricated revision.
+
 On driver entry, sweep under [Workspace hygiene](../axstack/references/workspace-hygiene.md); dispatched workers do not sweep.
 For every dispatch brief, name its private `<run dir>/evidence/<dispatch>/` folder.
 
@@ -64,7 +70,7 @@ Use actual records, never memory:
 - the decision log and configured `axstack-advisor-astra` and
   `axstack-advisor-opus` receipts and fresh `axstack-escalation-fable`
   receipts when triggered;
-- exact git revisions;
+- exact git revisions where applicable; owned SHA-256 snapshots for nonrepo evidence;
 - test and review evidence; and
 - the run execution record at its recorded `progress.md` path.
 

package/skills/axstack-audit/references/record.md

@@ -6,7 +6,7 @@ reason when evidence is missing; missing evidence is never a pass.
 
 ```text
 Run: <run identity + scope/authority + audit mode (end-of-run | checkpoint)>
-Baseline: <approved spec rev | peer mode | research mode | maintenance scope>
+Baseline: <approved spec rev | peer mode | research mode | maintenance scope | nonrepo report scope + source SHA-256>
 Acceptance: <passed / failed / unverified + test + SHA traces>
 Steps: <completed / deviated + why + approval per deviation>
 Advisers: <Astra/Opus configured coverage / eligible uses + same-question receipts; Astra/Fable high-stakes AGREE coverage / eligible uses>

package/skills/axstack-brainstorm/SKILL.md

@@ -10,6 +10,11 @@ Run inline in the current T3 driver thread. Never delegate a coordinator.
 This procedure is report-only. Do not implement, prototype, or grant execution
 or spec approval. Never interview the user or invoke Align.
 
+For an explicit No-project local report scope, follow
+[Local artifacts](../axstack/references/local-artifacts.md) for bounded candidate/judge
+roles, separate owned folders and SHA-256 receipts. Native null-identity proof,
+exact configured models/settings and arena round conditions remain required.
+
 Load [Standing contracts](../axstack/references/contracts.md) before acting.
 Set the rung from researched facts under the
 [design lens](../axstack/references/design-lens.md); Rung 2 meets its ADR test.

package/skills/axstack-brainstorm/references/arena.md

@@ -40,6 +40,14 @@ hard-to-reverse choices; they replace the critique round for that question.
    For candidate and judge threads, follow the [same-turn settle rule](../../axstack/references/workspace-hygiene.md#settlement).
    The user still decides; spec approval remains the one human checkpoint.
 
+For a No-project arena, use the explicit `workflow:brainstorm` scope in
+[Local artifacts](../../axstack/references/local-artifacts.md). Each candidate
+owns separate artifact/evidence roots; judges receive independently owned
+byte-identical anonymized snapshots after candidates finish and write only
+reports. Preserve exact SHA-256 readbacks and independent sessions, including
+fresh round-2 judges. Shared isolation grants no cross-reading or remote
+artifact permission. The scope allowlist does not waive rung/round triggers.
+
 Record the synthesis note (base, graft sources, rejections, dropouts, judge
 verdicts per round) as `Decisions` rows in the
 [caller's run record](../../axstack/references/run-record.md) when one exists,

package/skills/axstack-improve/SKILL.md

@@ -5,6 +5,12 @@ description: When assessing codebase quality, improving a named screen or flow,
 
 # Improve
 
+For named nonversioned code/UI discovery, use workflow improve under
+[Local artifacts](../axstack/references/local-artifacts.md). Its bounded code
+research/UI roles and inherited audit have separate owned snapshots and SHA-256
+receipts. This resolves the former routing ambiguity; it grants no repository
+readiness, whole-repository Scout, implementation or tracker authority.
+
 On driver entry, sweep under [Workspace hygiene](../axstack/references/workspace-hygiene.md); dispatched workers do not sweep.
 For every dispatch brief, name its private `<run dir>/evidence/<dispatch>/` folder.