fix: address the review findings raised on the v0.8.21 release PR

[waleedlatif1]

Summary

Fixes the review findings raised on the v0.8.21 release PR (#7446). Thirteen threads were triaged; each was verified against source before anything changed.

Three landed on staging independently while this was in progress — the /ship docs-manifest regeneration (#7450), the CLI Yarn upgrade tab (#7448), and the table view-pin (#7449) — so this branch was rebased and those duplicates dropped. What remains:

Knowledge / credential groups

• Slack could not use the connector Access field. It resolved the provider through the standard-OAuth subset, which excludes Slack (Slack collects accounts through a custom bot), so the field never rendered — Slack could not enter members mode, and a per-member Slack connector had no way back to workspace mode.
• The v1 document delete ran unscoped after an ACL-scoped lookup. The access-aware path already existed and v2 already used it; v1 was the last surface on the old one. The window is small and not attacker-controllable, but the divergence is worth closing.
• Enrollment surfaced CredentialGroupEnrollmentError as a bare 500. A missing, disabled, or unconfigured credential group is the admin's to act on, so its 404/409 now project the way the credential-group routes do.
• The workspace-level member-connector listing used knowledge-base concealment, answering Knowledge base not found where its siblings return an authorization response. It names a workspace, not a base — now unconcealed, matching the convention its own policy file documents.

Home

• A queued Build edit was stranded in Search. Restoring the mode left q in the URL, and the rule that forces Search whenever a query is present flipped the composer straight back: the edit was discarded and the original message dispatched. Clearing the query alongside the mode restore batches into the same nuqs update, so the forcing rule never observes the intermediate state.

Docs

• redis.mdx claimed completed work is unaffected by losing Redis. Webhook idempotency markers live wherever the cache does, with a 7-day TTL sized to the longest provider retry window, and the bundled Redis runs --save "" --appendonly no — so a redelivery after a restart can re-run a finished workflow with real side effects. Billing, checkout, and Chat-send idempotency are pinned to PostgreSQL. The same claim in values.yaml is corrected too.
• Documented that Ask may reach an integration for an explicitly requested action, not only when sources cannot answer.

Pushed back on (unchanged, reasoning left on each thread)

• custom-blocks "you enable" tracing — the FAQ addresses the publisher throughout, and the toggle is publisher-side (canManageBlock). The suggested rewrite would make the page inconsistent with every other "you" on it.
• Composer retains query after leaving Search — the defaultValue guard is deliberate (landing-prompt seed path). The suggested fix would destroy user-typed text on every mode switch and break the queued-edit fix above.
• Backtick inside a <source> payload — the file's own TSDoc documents this as an accepted trade; it only bites when the tag is also code-span-wrapped, costing one chip. Loosening the pattern removes the bound that stops a bogus match spanning code fences.

Type of Change

• Bug fix

Testing

bun run lint, bun run check:audits (45), bun run docs-manifest:check, and bun run type-check all pass; 1210 tests pass across lib/knowledge. The new access-scope test is mutation-verified — it goes red when the scope is not forwarded to the delete. /cleanup passes (effects, state, callbacks, url-state, comments), with its three comment findings applied.

Checklist

• Code follows project style guidelines
• Self-reviewed my changes
• Tests added/updated and passing
• No new warnings introduced
• I confirm that I have read and agree to the terms outlined in the Contributor License Agreement (CLA)

[vercel]

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

Project	Deployment	Actions	Updated
docs	Ready	Preview	Sep 3, 2026 9:22pm UTC

[greptile-apps]

Greptile Summary

The PR addresses review feedback from the v0.8.21 release work.

• Reapplies knowledge access scope when deleting documents and aligns knowledge-route error handling.
• Restores queued composer modes while clearing stale Search state.
• Enables Slack credential-group access options and excludes options that are not ready for provisioning.
• Corrects Redis persistence and webhook-idempotency documentation.
• Updates the Helm chart version.

Confidence Score: 5/5

The PR appears safe to merge.

No blocking failure remains.

Important Files Changed

Filename	Overview
apps/sim/lib/knowledge/orchestration/documents.ts	Replaces the unscoped document deletion with a knowledge-base- and access-scoped operation.
apps/sim/app/api/v1/knowledge/[id]/documents/[documentId]/route.ts	Resolves one access scope and forwards it through both document lookup and deletion.
apps/sim/app/workspace/[workspaceId]/home/home.tsx	Clears Search URL state before restoring the queued message's composer mode.
apps/sim/app/workspace/[workspaceId]/knowledge/[id]/hooks/use-connector-member-group-options.ts	Resolves all credential-group providers, including Slack, and exposes only provisionable options.
apps/sim/lib/knowledge/api/route-policies.ts	Projects credential-group enrollment failures and adds a workspace-level member-connector error policy.
apps/docs/content/docs/platform/self-hosting/redis.mdx	Documents the duplicate-side-effect risk when Redis-backed webhook idempotency markers are lost.
helm/sim/values.yaml	Clarifies Redis requirements and the consequences of running the bundled instance without persistence.

_{Reviews (3): Last reviewed commit: "fix: address the first review round" | Re-trigger Greptile}

[cubic-dev-ai]

All reported issues were addressed

<sub>Reply with feedback, questions, or to request a fix.

Fix all with cubic | Re-trigger cubic</sub>

[waleedlatif1]

@greptile

[waleedlatif1]

@cubic-dev-ai review this PR

[cubic-dev-ai]

@cubic-dev-ai review this PR

@waleedlatif1 I have started the AI code review. It will take a few minutes to complete.

[cubic-dev-ai]

No issues found across 11 files

Confidence score: 5/5

• Automated review surfaced no issues in the provided summaries.
• No files require special attention.

_{Re-trigger cubic}

[waleedlatif1]

@greptile

[waleedlatif1]

@cubic-dev-ai review this PR

[cubic-dev-ai]

@cubic-dev-ai review this PR

@waleedlatif1 I have started the AI code review. It will take a few minutes to complete.

[cubic-dev-ai]

2 issues found across 11 files

Confidence score: 3/5

• helm/sim/Chart.yaml leaves unpinned installs deploying v0.8.18 with the v0.8.21 release, creating a concrete default-install regression; set appVersion to v0.8.21 and regenerate the image inventory.
• helm/sim/values.yaml omits the Chat-send PostgreSQL idempotency exception from the Redis-loss guidance, which could lead operators to misunderstand chat-send deduplication behavior; document chatSendIdempotency alongside billing and checkout.

Prompt for AI agents (unresolved issues)

Check if these issues are valid — if so, understand the root cause of each and fix them. If appropriate, use sub-agents to investigate and fix each issue separately.


<file name="helm/sim/Chart.yaml">

<violation number="1" location="helm/sim/Chart.yaml:5">
P1: Unpinned chart installs still deploy `v0.8.18` even though this chart version is being shipped with the `v0.8.21` release. Set `appVersion` to `v0.8.21` and regenerate the image inventory so default installs receive the release features.

(Based on your team's feedback about keeping appVersion current for unpinned Helm installs.)</violation>
</file>

<file name="helm/sim/values.yaml">

<violation number="1" location="helm/sim/values.yaml:605">
P3: This Redis-loss note omits the `Chat-send` idempotency exception: `chatSendIdempotency` is forced to PostgreSQL, just like billing and checkout. Name it here so operators do not mistake chat-send deduplication for a Redis marker when evaluating restart risk.</violation>
</file>

<sub>Reply with feedback, questions, or to request a fix.

Fix all with cubic | Re-trigger cubic</sub>

[cubic-dev-ai]

P1: Unpinned chart installs still deploy v0.8.18 even though this chart version is being shipped with the v0.8.21 release. Set appVersion to v0.8.21 and regenerate the image inventory so default installs receive the release features.

(Based on your team's feedback about keeping appVersion current for unpinned Helm installs.)

View Feedback

Prompt for AI agents

Check if this issue is valid — if so, understand the root cause and fix it. At helm/sim/Chart.yaml, line 5:

<comment>Unpinned chart installs still deploy `v0.8.18` even though this chart version is being shipped with the `v0.8.21` release. Set `appVersion` to `v0.8.21` and regenerate the image inventory so default installs receive the release features.

(Based on your team's feedback about keeping appVersion current for unpinned Helm installs.) </comment>

<file context>
@@ -2,7 +2,7 @@ apiVersion: v2
 description: A Helm chart for Sim - the open-source AI workspace where teams build, deploy, and manage AI agents
 type: application
-version: 1.9.0
+version: 1.9.2
 appVersion: "v0.8.18"
 kubeVersion: ">=1.25.0-0"
</file context>

[cubic-dev-ai]

P3: This Redis-loss note omits the Chat-send idempotency exception: chatSendIdempotency is forced to PostgreSQL, just like billing and checkout. Name it here so operators do not mistake chat-send deduplication for a Redis marker when evaluating restart risk.

Prompt for AI agents

Check if this issue is valid — if so, understand the root cause and fix it. At helm/sim/values.yaml, line 605:

<comment>This Redis-loss note omits the `Chat-send` idempotency exception: `chatSendIdempotency` is forced to PostgreSQL, just like billing and checkout. Name it here so operators do not mistake chat-send deduplication for a Redis marker when evaluating restart risk.</comment>

<file context>
@@ -597,7 +599,10 @@ redis:
+  # keys, so a restart costs in-flight live updates, not committed data. It also
+  # drops webhook idempotency markers, so a provider redelivery after a
+  # restart can re-run an already-completed workflow — use a persistent managed
+  # instance if that matters. Billing and checkout idempotency is on PostgreSQL.
   maxmemory: "512mb"
   maxmemoryPolicy: "noeviction"
</file context>

Suggested change

	# instance if that matters. Billing and checkout idempotency is on PostgreSQL.
	# instance if that matters. Billing, checkout, and Chat-send idempotency are on PostgreSQL.