Conversation
…PC handshake The protobufjs override to ^7.5.5 (resolves 7.6.6) has a breaking change with @grpc/[email protected] — gRPC handshake fails (error 14 UNAVAILABLE, 'Failed to connect before the deadline') even though TCP connects. This breaks ALL Temporal workflow creation (startWorkflow silently fails via D6 pattern).
…DK and MCP The branding guard only scanned for AGPL attribution (repo, image, author), so it reported "0 strict leaks" while runtime endpoints still pointed at upstream infrastructure: - the Developers page rendered `export POSTIZ_API_KEY=<real key>` with no POSTIZ_API_URL, so the upstream CLI sent Crove credentials to api.postiz.com and posted to the wrong account - `postiz auth:login` ran a device flow against cli-auth.postiz.com, signing customers into the upstream cloud - the rebranded @crove/node SDK still defaulted _path to api.postiz.com, and its README still documented the old @postiz/node package name - the MCP server announced itself as "Postiz MCP" on every OAuth path, including /mcp-oauth-dynamic which DOSClaw uses - "Add to Claude" pointed at the upstream directory listing and the Affiliate menu item at affiliate.postiz.com Guard changes: add a RUNTIME pattern family alongside the existing AGPL attribution patterns, bring apps/sdk/ into STRICT scope, drop the now-stale apps/sdk/package.json exception, report per-match line numbers instead of attributing every hit in a file to the first line, and add a `branding-guard-allow:` sentinel for references that are legitimately upstream. Allowed hits are still counted and printed so exceptions stay visible rather than silently accumulating. Add BRAND_CLAUDE_DIRECTORY_URL, defaulting to empty so the Add-to-Claude button fails closed instead of offering the upstream listing. Collapse the CLI section's Locally/CI toggle into one list: with no self-hosted CLI auth server both variants reduce to the same API-key flow. Visible behaviour changes, all fail-closed: the Add-to-Claude button and the Affiliate menu item are hidden until BRAND_CLAUDE_DIRECTORY_URL and BRAND_AFFILIATE_URL are set. Docs links now resolve through BRAND_DOCS_URL, which is still configured as docs.postiz.com in the deployment env files and needs a separate decision. Also fix docs/upstream-sync.md: the documented `pnpm exec tsx` invocation fails because tsx is not a dependency, and the workflow triggers on main/dev, not main/master.
… of hardcoding it
Every generated client snippet registered this server under the literal key
"postiz" — `claude mcp add postiz`, `{ mcpServers: { postiz: … } }`,
`[mcp_servers.postiz]`, `cursor://…?name=postiz` — so customers saw an
upstream-branded connector inside their own Claude/Cursor/Codex.
Add BrandConfig.mcpConnectorName, derived from BRAND_SHORT_NAME and
overridable via BRAND_MCP_CONNECTOR_NAME. It is always slugified so it stays
safe as a shell argument, JSON/YAML key, TOML table name and URL query param;
prod resolves to "crove" and beta to "crove-beta", so the two environments no
longer collide in the same client. Verified by rendering every snippet
against the real deployment env files.
Add joinBrandUrl and use it for every docs link. sanitizeUrl normalises a
bare origin to a trailing slash, so the previous `${docsUrl}/path`
concatenation emitted a double slash — a regression introduced while
de-hardcoding those URLs in the previous commit.
Extend the guard with patterns for connector keys baked into generated
config, and record the three identifiers that must stay upstream: the compose
service name (container_name, the postiz-postgres dependency, the volume
names and validate-beta-compose.mjs all key off it) and the two
backward-compatible `postiz` agent aliases documented in
docs/architecture.md §8.2.
Also rename the SDK's default export Postiz -> Crove to match the
@crove/node package name and its README.
…on in Node 22 Fixes Facebook post publishing failure: @grpc/[email protected] resolves ALL hostnames to IPv6 ::1 in Node 22 Docker containers, causing ECONNREFUSED for every Temporal workflow start. Downgrading to 1.12.4 restores working DNS resolution.
Bugbot couldn't run - usage limit reachedBugbot is counted against Cursor usage for this user or team, and this run hit a usage or spend limit. A user or team admin can review and increase usage limits in the Cursor dashboard. (requestId: serverGenReqId_d328621f-be3d-4d14-972c-696e3b393d1c) |
![gemini-code-assist[bot]](https://avatars.githubusercontent.com/in/956858?s=60&v=4){kind=small}
There was a problem hiding this comment.
Code Review
This pull request introduces extensive rebranding and white-labeling improvements to decouple the application from the upstream "Postiz" brand and configure it for "Crove". Key changes include replacing hardcoded upstream URLs (documentation, affiliate, and Claude directory links) with dynamic, brand-configurable values, introducing helper functions (slugifyIdentifier and joinBrandUrl) to handle brand configurations safely, and updating the onboarding and developer components to dynamically generate MCP and CLI configurations using a customizable mcpConnectorName. Additionally, the NodeJS SDK has been rebranded to Crove, and the branding-guard.ts script has been enhanced with stricter checks to prevent runtime leaks of upstream endpoints and identifiers. I have no feedback to provide as there are no review comments.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
What kind of change does this PR introduce?
Bug fix — scope: dependencies (
package.json,pnpm-lock.yaml). Fixes the Facebook post publishing failure caused by@grpc/[email protected]resolving ALL hostnames to IPv6::1in Node 22 Docker containers.Why was this change needed?
After deploying the new image (PR #34), ALL Temporal workflow starts silently failed. The Temporal client returned gRPC error 14 (UNAVAILABLE) with
ECONNREFUSED ::1:7233— the gRPC library resolves every hostname (including IP addresses like172.18.0.8) to IPv6 loopback regardless of/etc/hosts,dns-result-order=ipv4first, or theipv4://URI scheme. This caused the user's Facebook post to never publish (D6 silent failure pattern).Technical Details & Scope
@grpc/grpc-jsdowngraded from 1.14.3 to 1.12.4 viapnpm.overrides::1for all hostnames (even direct IP addresses)protobufjsoverride removed (the^7.5.5override resolved to 7.6.6 which also broke gRPC handshake)Verification & Testing
pnpm --filter ./apps/backend run build— exit 0pnpm --filter ./apps/orchestrator run build— exit 0QA
pnpm list @grpc/grpc-js --depth 0— expect 1.12.4pnpm --filter ./apps/backend run build— expect exit 0docker exec crove-post node -e "const{Client}=require('@temporalio/client');const c=new Client({address:process.env.TEMPORAL_ADDRESS});c.connection.workflowService.describeNamespace({namespace:'default'}).then(r=>{console.log('GRPC_OK');process.exit(0)}).catch(e=>{console.log('FAIL',e.code);process.exit(1)})" | tail -1— expect GRPC_OKChecklist:
pnpm run build).pnpm dlx tsx scripts/branding-guard.ts).Note
Medium Risk
Touches customer-facing integration instructions and API/CLI targeting (misconfiguration risk), plus a pinned gRPC downgrade that affects Temporal workflow connectivity.
Overview
Extends the brand config with
mcpConnectorName(fromBRAND_MCP_CONNECTOR_NAMEor slugified short name),claudeDirectoryUrl, andjoinBrandUrlso developer/onboarding/public-API UI stop hardcoding Postiz docs, affiliate, and Claude directory links. MCP snippets, Cursor install links, and onboarding now use the branded connector key instead of a literalpostiz; Add to Claude and the affiliate menu only appear when the corresponding brand URLs are set (empty defaults hide upstream listings).CLI guidance is unified around API-key auth for self-hosted: mandatory
POSTIZ_API_URLfor this instance, removal of the localpostiz auth:login/ CI toggle, and chat-only agent copy that sets both URL and key. The Node SDK/README default base moves to Crove’s API path; MCP OAuth server display names use${brand.name} MCP.Branding guard gains tests for the new fields, runtime leak patterns (endpoints, connector keys, Claude directory), stricter scanning under
apps/sdk/, andbranding-guard-allowexceptions. Dependencies:@grpc/grpc-jspinned to 1.12.4 andprotobufjsto 7.5.5 in overrides/lockfile (Temporal gRPC on Node 22).Reviewed by Cursor Bugbot for commit f8ff880. Configure here.