v0.9.11: managed mcp optimizations, instrumentation improvements, mammoth parser bump

[waleedlatif1]

• improvement(audits): enforce console, helper, render-path, persist, and deployment-flag rules (improvement(audits): enforce console, helper, render-path, persist, and deployment-flag rules #8559)
• docs(agents): accurate guidance, guardrail map, and check:guidance-refs (docs(agents): accurate guidance, guardrail map, and check:guidance-refs #8557)
• fix(search): reuse managed MCP sessions within operations (fix(search): reuse managed MCP sessions within operations #8567)
• fix(files): resolve chat uploads whose names are not in VFS form (fix(files): resolve chat uploads whose names are not in VFS form #8566)
• fix(logging): name the driver cause and redact bound params in logged errors (fix(logging): name the driver cause and redact bound params in logged errors #8565)
• fix(deps): bump mammoth to 1.12.3 (fix(deps): bump mammoth to 1.12.3 #8568)

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated
docs	Ready	Preview	Oct 2, 2026 3:49pm UTC

[greptile-apps]

Confidence Score: 4/5

[Medium risk] Documentation and configuration updates across agent skills and rules.

The PR should not merge until normalized chat-upload names resolve unambiguously.

Findings

1.  Upload references can select wrong file ▶

Summary

The PR refreshes agent guidance and audit checks, reuses managed MCP sessions within Search operations, improves chat-upload name resolution and logging, and bumps mammoth. One upload-resolution collision can cause a chat file reference to return the wrong file.

Diagram

%%{init: {'theme': 'neutral'}}%%
flowchart LR
  A["Two distinct stored chat names"] --> B["Same normalized uploads path"]
  B --> C["Normalized name lookup"]
  C --> D["Newest matching upload returned"]

Loading

_{Reviews (1) · Last reviewed commit: "fix(deps): bump mammoth to 1.12.3 (#8568..."}

[cubic-dev-ai]

15 issues found across 211 files

Confidence score: 3/5

• managed-mcp.integration.ts returns SECOND_DOCUMENT from search and metadata but fetches a manifest identified as DOCUMENT, so the Lucid reader rejects the second-document integration cases. Make the fetched manifest identity match the searched document.
• check-zustand-v5-selectors.ts can miss persist calls without a partialize callback and whole-state persistence through destructured rest parameters. Check the persist options for partialize and reject rest-based whole-state selectors.
• check-application-graph.ts treats integration tests and declaration files as live modules, so they can mask a deleted or renamed guarded runtime module. Exclude those files from the live-module match.
• check-client-boundary-imports.ts misses namespace members read through destructuring, such as const { isHosted } = flags. Detect destructured members so this boundary check cannot be bypassed.

Prompt for AI agents (unresolved issues)

Check if these issues are valid — if so, understand the root cause of each and fix them. When an issue isn't valid or won't be fixed in this PR, reply in its thread with the reason and then resolve the thread. If appropriate, use sub-agents to investigate and fix each issue separately.


<file name="apps/sim/lib/sim-search/live/managed-mcp.integration.ts">

<violation number="1" location="apps/sim/lib/sim-search/live/managed-mcp.integration.ts:120">
P2: The fixture returns `SECOND_DOCUMENT` from search and metadata but identifies every fetched manifest as `DOCUMENT`; the Lucid reader rejects that identity mismatch, causing the second-document integration cases to fail. Derive the fetch manifest identity from the requested `id`.</violation>
</file>

<file name=".claude/rules/sim-url-state.md">

<violation number="1" location=".claude/rules/sim-url-state.md:17">
P3: Condensing the decision bullets into the table dropped the only reference to `.claude/rules/sim-queries.md` in this file (the removed "React Query → server/remote data. Unchanged; see sim-queries.md" bullet). The doc set keeps reciprocal pointers between these rules (sim-queries.md:11 points back to sim-url-state.md, sim-api-contracts.md and sim-hooks.md point at sim-queries.md), so readers/agents now lose the route to the authoritative React Query rule. Re-add the cross-reference, e.g. in the React Query row or the intro sentence.</violation>
</file>

<file name=".agents/skills/add-block-preview/SKILL.md">

<violation number="1" location=".agents/skills/add-block-preview/SKILL.md:56">
P3: `lib/integrations/tool-catalog.ts` is not resolvable from the repo root where this doc lives; the file is at `apps/sim/lib/integrations/tool-catalog.ts` (imported in code as `@/lib/integrations/tool-catalog.ts`). Every sibling reference in this file uses the full `apps/sim/...` prefix — align the path so an agent following the guidance can find the file.</violation>
</file>

<file name=".cursor/rules/sim-styling.mdc">

<violation number="1" location=".cursor/rules/sim-styling.mdc:98">
P3: The `See .../usage-limit-field.tsx` reference no longer demonstrates the `inputClassName` example: the file uses `<ChipInput inputMode='numeric' />` and contains no `inputClassName`, `font-mono`, or number-spinner reset (also true at pr-base). Drop the pointer or update it to a file that actually uses the prop, otherwise the example it is meant to illustrate is not there.</violation>
</file>

<file name=".agents/skills/babysit/SKILL.md">

<violation number="1" location=".agents/skills/babysit/SKILL.md:91">
P2: Step 6's sync check only rebases when `git log origin/staging..HEAD` shows unrecognized commits (`/ship` step 2: "If it shows commits you don't recognize, fix it now... Try `git rebase origin/staging` first"). A PR that conflicts purely because staging advanced carries only this session's recognizable commits, so following this pointer to step 6 runs no rebase and the conflict is never resolved — step 7 then has nothing new to push and step 8 re-triggers reviewers on a still-`CONFLICTING` branch, repeating until step 10's two-round stop condition surfaces it to the user. State the resolution action explicitly: rebase onto `origin/staging`, resolve, and `git rebase --continue` before running the gates.</violation>
</file>

<file name=".cursor/rules/sim-queries.mdc">

<violation number="1" location=".cursor/rules/sim-queries.mdc:144">
P2: Removing this `eslint-disable-next-line react-hooks/exhaustive-deps` leaves the "✓ Good" example failing the lint rule it documents: `useCallback` referencing `createEntity` with deps `[data]` reports a missing dependency. The repo's own usage of this exact pattern (e.g. `apps/sim/app/workspace/[workspaceId]/tables/tables.tsx` uses `// eslint-disable-next-line react-hooks/exhaustive-deps -- mutation objects are unstable; mutate is stable in v5`) shows the disable comment is required, so the doc now presents code that cannot be committed verbatim cleanly. Restore the comment (with the explanatory annotation) or address the lint trigger explicitly.</violation>
</file>

<file name=".agents/skills/add-settings-page/SKILL.md">

<violation number="1" location=".agents/skills/add-settings-page/SKILL.md:50">
P3: The listed expected match `CredentialDetailLayout` can never appear in this grep's results: credential-detail-layout.tsx lives at workspace `components/credential-detail/...`, outside both `'apps/sim/**/settings/**'` and `'apps/sim/ee/'`. Drop it from the expected-match list (or widen the pathspec to actually include it), and rephrase so it doesn't clash with 'A detail sub-view is never a match' — the point is that it is exempt because it hand-rolls the shell, not because it is a grep match.</violation>
</file>

<file name=".claude/rules/sim-settings-pages.md">

<violation number="1" location=".claude/rules/sim-settings-pages.md:96">
P3: Not every `SETTINGS_SECTION_REGISTRY` entry carries a `description`: the field lives on the optional `unified` projection (`SettingsSectionRegistryEntry` only declares `label`, `icon`, `docsLink`, `unified?`, `planes?`), and two entries (`Chat keys`, `Recently deleted`) have no `unified` at all. The next hunk of this same change already uses the correct contract (`label` and `unified.description`); align this sentence with it so agents don't force a `unified.description` onto standalone-plane entries.</violation>
</file>

<file name="packages/logger/src/index.ts">

<violation number="1" location="packages/logger/src/index.ts:218">
P2: The reverse scan makes OTel attributes select the last bare error, so calls with multiple errors export the wrong primary exception. Iterate `args` in order while retaining the existing `{ error }` fallback.

(Based on your team's feedback about OTel primary-error selection.)</violation>
</file>

<file name="scripts/check-application-graph.ts">

<violation number="1" location="scripts/check-application-graph.ts:298">
P2: `prefixMatchesAnything` treats `*.integration.ts` and declaration files as live modules. A renamed or deleted guarded runtime module can therefore leave only test or type files and still bypass the dead-prefix failure; exclude those suffixes alongside `.test.ts`.</violation>
</file>

<file name="scripts/check-zustand-v5-selectors.ts">

<violation number="1" location="scripts/check-zustand-v5-selectors.ts:319">
P2: `hasPartialize` can be satisfied by a state member or storage name, allowing a persist call without a `partialize` callback to bypass the audit. Check the second/options argument for the `partialize` property before treating the call as compliant.</violation>

<violation number="2" location="scripts/check-zustand-v5-selectors.ts:322">
P2: The whole-state check misses destructured rest parameters, so `partialize: ({ ...state }) => ({ ...state })` and its block-body equivalent pass despite persisting every field and action. Extend the check to reject destructured-rest callbacks that return the rest object.

(Based on your team's feedback about rejecting whole-state partializers.)</violation>
</file>

<file name="scripts/check-client-boundary-imports.ts">

<violation number="1" location="scripts/check-client-boundary-imports.ts:106">
P2: Namespace destructuring bypasses this check: `envFlagReads` only recognizes `flags.name`, so `const { isHosted } = flags` is never reported. Detect destructured namespace members too.</violation>
</file>

<file name="scripts/check-utils-enforcement.ts">

<violation number="1" location="scripts/check-utils-enforcement.ts:149">
P2: The `.with` check only matches numeric literal indexes, so `items.with(index, nextValue)` bypasses this array-method ban. Detect variable-index array calls without flagging non-array `context.with()` methods.</violation>
</file>

<file name=".agents/skills/add-integration/SKILL.md">

<violation number="1" location=".agents/skills/add-integration/SKILL.md:135">
P3: This points authors to conflicting canonical-ID rules: the `add-block` skill forbids matching any subblock ID, while this guidance and `sim-integrations.md` allow a group member to share it. Synchronize the referenced `add-block` rule so authors do not reject valid pairs or follow the wrong constraint.</violation>
</file>

<sub>Note: This PR contains a large number of files. cubic selects up to 200 of the highest-priority eligible files for this review, so some files may not have been reviewed.
Tip: instead of fixing issues one by one fix them all with cubic

Re-trigger cubic</sub>

[cubic-dev-ai]

P2: The fixture returns SECOND_DOCUMENT from search and metadata but identifies every fetched manifest as DOCUMENT; the Lucid reader rejects that identity mismatch, causing the second-document integration cases to fail. Derive the fetch manifest identity from the requested id.

Prompt for AI agents

Check if this issue is valid — if so, understand the root cause and fix it. When an issue isn't valid or won't be fixed in this PR, reply in its thread with the reason and then resolve the thread. At apps/sim/lib/sim-search/live/managed-mcp.integration.ts, line 120:

<comment>The fixture returns `SECOND_DOCUMENT` from search and metadata but identifies every fetched manifest as `DOCUMENT`; the Lucid reader rejects that identity mismatch, causing the second-document integration cases to fail. Derive the fetch manifest identity from the requested `id`.</comment>

<file context>
@@ -0,0 +1,677 @@
+      protocol.setRequestHandler(CallToolRequestSchema, async ({ params }) => {
+        events.push({ method: params.name, actor, at: performance.now() })
+        await onTool?.(params.name, params.arguments ?? {})
+        const documentId =
+          params.arguments?.query === 'second topology' ||
+          params.arguments?.document_id === SECOND_DOCUMENT
</file context>

[cubic-dev-ai]

P2: Step 6's sync check only rebases when git log origin/staging..HEAD shows unrecognized commits (/ship step 2: "If it shows commits you don't recognize, fix it now... Try git rebase origin/staging first"). A PR that conflicts purely because staging advanced carries only this session's recognizable commits, so following this pointer to step 6 runs no rebase and the conflict is never resolved — step 7 then has nothing new to push and step 8 re-triggers reviewers on a still-CONFLICTING branch, repeating until step 10's two-round stop condition surfaces it to the user. State the resolution action explicitly: rebase onto origin/staging, resolve, and git rebase --continue before running the gates.

Prompt for AI agents

Check if this issue is valid — if so, understand the root cause and fix it. When an issue isn't valid or won't be fixed in this PR, reply in its thread with the reason and then resolve the thread. At .agents/skills/babysit/SKILL.md, line 91:

<comment>Step 6's sync check only rebases when `git log origin/staging..HEAD` shows unrecognized commits (`/ship` step 2: "If it shows commits you don't recognize, fix it now... Try `git rebase origin/staging` first"). A PR that conflicts purely because staging advanced carries only this session's recognizable commits, so following this pointer to step 6 runs no rebase and the conflict is never resolved — step 7 then has nothing new to push and step 8 re-triggers reviewers on a still-`CONFLICTING` branch, repeating until step 10's two-round stop condition surfaces it to the user. State the resolution action explicitly: rebase onto `origin/staging`, resolve, and `git rebase --continue` before running the gates.</comment>

<file context>
@@ -88,8 +88,9 @@ conditions freshly after every push.
 
-2. **If the PR has a merge conflict**, merge `origin/staging`, resolve the conflicts, run the
-   usual pre-push checks, push, and go to step 8 to re-trigger review.
+2. **If the PR has a merge conflict**, resolve it with step 6 (its rebase-based sync flow and
+   the `/ship` gates; a merge commit would be discarded by that rebase), then steps 7–8: push
+   with `--force-with-lease` and re-trigger review.
</file context>

Suggested change

	2. **If the PR has a merge conflict**, resolve it with step 6 (its rebase-based sync flow and
	the `/ship` gates; a merge commit would be discarded by that rebase), then steps 7–8: push
	with `--force-with-lease` and re-trigger review.
	2. **If the PR has a merge conflict**, run `git rebase origin/staging`, resolve the conflicts, and
	`git rebase --continue` (this is step 6's rebase-based sync flow — a merge commit would be
	discarded by that rebase), then run step 6's `/ship` gates and steps 7–8: push with
	`--force-with-lease` and re-trigger review.

[cubic-dev-ai]

P2: Removing this eslint-disable-next-line react-hooks/exhaustive-deps leaves the "✓ Good" example failing the lint rule it documents: useCallback referencing createEntity with deps [data] reports a missing dependency. The repo's own usage of this exact pattern (e.g. apps/sim/app/workspace/[workspaceId]/tables/tables.tsx uses // eslint-disable-next-line react-hooks/exhaustive-deps -- mutation objects are unstable; mutate is stable in v5) shows the disable comment is required, so the doc now presents code that cannot be committed verbatim cleanly. Restore the comment (with the explanatory annotation) or address the lint trigger explicitly.

Prompt for AI agents

Check if this issue is valid — if so, understand the root cause and fix it. When an issue isn't valid or won't be fixed in this PR, reply in its thread with the reason and then resolve the thread. At .cursor/rules/sim-queries.mdc, line 144:

<comment>Removing this `eslint-disable-next-line react-hooks/exhaustive-deps` leaves the "✓ Good" example failing the lint rule it documents: `useCallback` referencing `createEntity` with deps `[data]` reports a missing dependency. The repo's own usage of this exact pattern (e.g. `apps/sim/app/workspace/[workspaceId]/tables/tables.tsx` uses `// eslint-disable-next-line react-hooks/exhaustive-deps -- mutation objects are unstable; mutate is stable in v5`) shows the disable comment is required, so the doc now presents code that cannot be committed verbatim cleanly. Restore the comment (with the explanatory annotation) or address the lint trigger explicitly.</comment>

<file context>
@@ -141,7 +141,6 @@ const handler = useCallback(() => {
 const handler = useCallback(() => {
   createEntity.mutate(data)
-  // eslint-disable-next-line react-hooks/exhaustive-deps
 }, [data])

</file context>

</details>

<a href="https://www.cubic.dev/action/fix/violation/5a275cb7-1503-4197-8c7c-5c672cc85d35" target="_blank" rel="noopener noreferrer" data-no-image-dialog="true">
  <picture>
    <source media="(prefers-color-scheme: dark)" srcset="https://cubic.dev/buttons/fix-with-cubic-dark.svg">
    <source media="(prefers-color-scheme: light)" srcset="https://cubic.dev/buttons/fix-with-cubic-light.svg">
    <img alt="Fix with cubic" src="https://cubic.dev/buttons/fix-with-cubic-dark.svg">
  </picture>
</a>

[cubic-dev-ai]

P2: The reverse scan makes OTel attributes select the last bare error, so calls with multiple errors export the wrong primary exception. Iterate args in order while retaining the existing { error } fallback.

(Based on your team's feedback about OTel primary-error selection.)

View Feedback

Prompt for AI agents

Check if this issue is valid — if so, understand the root cause and fix it. When an issue isn't valid or won't be fixed in this PR, reply in its thread with the reason and then resolve the thread. At packages/logger/src/index.ts, line 218:

<comment>The reverse scan makes OTel attributes select the last bare error, so calls with multiple errors export the wrong primary exception. Iterate `args` in order while retaining the existing `{ error }` fallback.

(Based on your team's feedback about OTel primary-error selection.) </comment>

<file context>
@@ -131,55 +132,105 @@ const getLogConfig = () => {
+ * `Error` argument, else the first `{ error }` field.
+ */
+const primaryError = (args: unknown[]): Error | undefined => {
+  for (let i = args.length - 1; i >= 0; i--) {
+    const arg = args[i]
+    if (arg instanceof Error) return arg
</file context>

Suggested change

	for (let i = args.length - 1; i >= 0; i--) {
	const arg = args[i]
	if (arg instanceof Error) return arg
	}
	for (const arg of args) {
	if (arg instanceof Error) return arg
	}

[cubic-dev-ai]

P3: lib/integrations/tool-catalog.ts is not resolvable from the repo root where this doc lives; the file is at apps/sim/lib/integrations/tool-catalog.ts (imported in code as @/lib/integrations/tool-catalog.ts). Every sibling reference in this file uses the full apps/sim/... prefix — align the path so an agent following the guidance can find the file.

Prompt for AI agents

Check if this issue is valid — if so, understand the root cause and fix it. When an issue isn't valid or won't be fixed in this PR, reply in its thread with the reason and then resolve the thread. At .agents/skills/add-block-preview/SKILL.md, line 56:

<comment>`lib/integrations/tool-catalog.ts` is not resolvable from the repo root where this doc lives; the file is at `apps/sim/lib/integrations/tool-catalog.ts` (imported in code as `@/lib/integrations/tool-catalog.ts`). Every sibling reference in this file uses the full `apps/sim/...` prefix — align the path so an agent following the guidance can find the file.</comment>

<file context>
@@ -53,7 +53,7 @@ To pull an already-GA block from discovery surfaces on hosted (incident, depreca
 - **Keys are registry block types.** Never `custom_block_*` (parse drops them — custom blocks have their own enabled/disabled lifecycle).
 - **The shared hidden-predicate is `isHiddenUnder`** (`apps/sim/blocks/visibility/context.ts`). Never restate the preview/disabled rule inline at a new consumer.
-- **Process-global caches stay ungated.** `getStaticComponentFiles` (VFS) and `getExposedIntegrationTools` build the ungated universe; per-viewer filtering happens at stamp/consumer time. Never move gating into a shared builder.
+- **Process-global caches stay ungated.** Shared builders such as `getExposedIntegrationTools` (`lib/integrations/tool-catalog.ts`) build the ungated universe; per-viewer filtering happens at consumer time via `isHiddenUnder`. Never move gating into a shared builder.
 - Gating is **surface hiding, not secrecy** — the full config ships in the client JS bundle. Anything truly secret cannot be a registered block.
 
</file context>

Suggested change

	- **Process-global caches stay ungated.** Shared builders such as `getExposedIntegrationTools` (`lib/integrations/tool-catalog.ts`) build the ungated universe; per-viewer filtering happens at consumer time via `isHiddenUnder`. Never move gating into a shared builder.
	**Process-global caches stay ungated.** Shared builders such as `getExposedIntegrationTools` (`apps/sim/lib/integrations/tool-catalog.ts`) build the ungated universe; per-viewer filtering happens at consumer time via `isHiddenUnder`. Never move gating into a shared builder.

[cubic-dev-ai]

P3: The See .../usage-limit-field.tsx reference no longer demonstrates the inputClassName example: the file uses <ChipInput inputMode='numeric' /> and contains no inputClassName, font-mono, or number-spinner reset (also true at pr-base). Drop the pointer or update it to a file that actually uses the prop, otherwise the example it is meant to illustrate is not there.

Prompt for AI agents

Check if this issue is valid — if so, understand the root cause and fix it. When an issue isn't valid or won't be fixed in this PR, reply in its thread with the reason and then resolve the thread. At .cursor/rules/sim-styling.mdc, line 98:

<comment>The `See .../usage-limit-field.tsx` reference no longer demonstrates the `inputClassName` example: the file uses `<ChipInput inputMode='numeric' />` and contains no `inputClassName`, `font-mono`, or number-spinner reset (also true at pr-base). Drop the pointer or update it to a file that actually uses the prop, otherwise the example it is meant to illustrate is not there.</comment>

<file context>
@@ -95,19 +95,19 @@ Draw a line with a real `border-*` utility. Never hand-roll one as `shadow-[inse
 - **Trailing buttons** (reveal/copy/fetch) → `endAdornment`.
-- **Inner-input styling** (e.g. `font-mono`, number-spinner reset) → `inputClassName` (ChipInput only). See `app/workspace/[workspaceId]/settings/components/billing/components/usage-limit-field/usage-limit-field.tsx:117`.
-- **`ChipModalField` controls take NO className.** Pass `title`/`value`/`onChange`/`error`/`hint`/`required`/`flush`. The field owns label, control, and error/hint rendering. See `app/workspace/[workspaceId]/skills/components/skill-modal/skill-modal.tsx:170`.
+- **Inner-input styling** (e.g. `font-mono`, number-spinner reset) → `inputClassName` (ChipInput only). See `app/workspace/[workspaceId]/settings/components/billing/components/usage-limit-field/usage-limit-field.tsx`.
+- **`ChipModalField` controls take NO className.** Pass `title`/`value`/`onChange`/`error`/`hint`/`required`/`flush`. The field owns label, control, and error/hint rendering. See `app/workspace/[workspaceId]/skills/components/skill-modal/skill-modal.tsx`.
 
</file context>

Suggested change

	- **Inner-input styling** (e.g. `font-mono`, number-spinner reset) → `inputClassName` (ChipInput only). See `app/workspace/[workspaceId]/settings/components/billing/components/usage-limit-field/usage-limit-field.tsx`.
	- **Inner-input styling** (e.g. `font-mono`, number-spinner reset) → `inputClassName` (ChipInput only).

[cubic-dev-ai]

P3: The listed expected match CredentialDetailLayout can never appear in this grep's results: credential-detail-layout.tsx lives at workspace components/credential-detail/..., outside both 'apps/sim/**/settings/**' and 'apps/sim/ee/'. Drop it from the expected-match list (or widen the pathspec to actually include it), and rephrase so it doesn't clash with 'A detail sub-view is never a match' — the point is that it is exempt because it hand-rolls the shell, not because it is a grep match.

Prompt for AI agents

Check if this issue is valid — if so, understand the root cause and fix it. When an issue isn't valid or won't be fixed in this PR, reply in its thread with the reason and then resolve the thread. At .agents/skills/add-settings-page/SKILL.md, line 50:

<comment>The listed expected match `CredentialDetailLayout` can never appear in this grep's results: credential-detail-layout.tsx lives at workspace `components/credential-detail/...`, outside both `'apps/sim/**/settings/**'` and `'apps/sim/ee/'`. Drop it from the expected-match list (or widen the pathspec to actually include it), and rephrase so it doesn't clash with 'A detail sub-view is never a match' — the point is that it is exempt because it hand-rolls the shell, not because it is a grep match.</comment>

<file context>
@@ -21,65 +21,58 @@ Key paths:
-4. Confirm each page imports `SettingsPanel` and that its `NavigationItem` has an
+   `git grep -n "flex h-full flex-col bg-\[var(--bg)\]" -- 'apps/sim/**/settings/**' 'apps/sim/ee/'`
+   — expected matches: the workspace and organization `settings/layout.tsx` shells, the shared
+   header shell (`components/settings/settings-header.tsx`), `CredentialDetailLayout` (the
+   `settings/secrets/[credentialId]` exception), or an entitlement/loading gate. A detail
+   sub-view is never a match: it passes `back={{ text, icon: ArrowLeft, onSelect }}` to
</file context>

[cubic-dev-ai]

P3: Not every SETTINGS_SECTION_REGISTRY entry carries a description: the field lives on the optional unified projection (SettingsSectionRegistryEntry only declares label, icon, docsLink, unified?, planes?), and two entries (Chat keys, Recently deleted) have no unified at all. The next hunk of this same change already uses the correct contract (label and unified.description); align this sentence with it so agents don't force a unified.description onto standalone-plane entries.

Prompt for AI agents

Check if this issue is valid — if so, understand the root cause and fix it. When an issue isn't valid or won't be fixed in this PR, reply in its thread with the reason and then resolve the thread. At .claude/rules/sim-settings-pages.md, line 96:

<comment>Not every `SETTINGS_SECTION_REGISTRY` entry carries a `description`: the field lives on the optional `unified` projection (`SettingsSectionRegistryEntry` only declares `label`, `icon`, `docsLink`, `unified?`, `planes?`), and two entries (`Chat keys`, `Recently deleted`) have no `unified` at all. The next hunk of this same change already uses the correct contract (`label` and `unified.description`); align this sentence with it so agents don't force a `unified.description` onto standalone-plane entries.</comment>

<file context>
@@ -93,17 +93,17 @@ return (
 
 `apps/sim/components/settings/navigation.ts` is the single source of truth (the
-`settings/navigation.ts` in the route tree is only a re-export shim). Every `NavigationItem` carries a one-line `description`; `SettingsPanel`
+`settings/navigation.ts` in the route tree is only a re-export shim). Every `SETTINGS_SECTION_REGISTRY` entry carries a one-line `description`; `SettingsPanel`
 resolves both via `getSettingsSectionMeta(plane, section)` and the
 `SettingsSectionProvider` the settings shell wraps around the active section.
</file context>

Suggested change

	`settings/navigation.ts` in the route tree is only a re-export shim). Every `SETTINGS_SECTION_REGISTRY` entry carries a one-line `description`; `SettingsPanel`
	`settings/navigation.ts` in the route tree is only a re-export shim). Every `SETTINGS_SECTION_REGISTRY` entry with a `unified` projection carries a one-line `unified.description`; `SettingsPanel`

[cubic-dev-ai]

P3: This points authors to conflicting canonical-ID rules: the add-block skill forbids matching any subblock ID, while this guidance and sim-integrations.md allow a group member to share it. Synchronize the referenced add-block rule so authors do not reject valid pairs or follow the wrong constraint.

Prompt for AI agents

Check if this issue is valid — if so, understand the root cause and fix it. When an issue isn't valid or won't be fixed in this PR, reply in its thread with the reason and then resolve the thread. At .agents/skills/add-integration/SKILL.md, line 135:

<comment>This points authors to conflicting canonical-ID rules: the `add-block` skill forbids matching any subblock ID, while this guidance and `sim-integrations.md` allow a group member to share it. Synchronize the referenced `add-block` rule so authors do not reject valid pairs or follow the wrong constraint.</comment>

<file context>
@@ -128,13 +128,11 @@ Three rules that are easy to get wrong when copying from existing blocks:
-  survives serialization, so `inputs` and `tools.config.params` reference the canonical id, never the
-  subblock ids. It is unique block-wide, and every member of a group shares the same `required` value.
+- Basic/advanced pairs use a `canonicalParamId`; its constraints are in
+  `.claude/rules/sim-integrations.md` and the `add-block` skill → canonicalParamId Pattern.
 - Every text-entry subBlock (`short-input`, `long-input`, `code`) and every selector declares a
   `placeholder`; an empty box tells the user nothing. Secrets read `Enter your {thing}` (e.g.
</file context>