Summary
Server detail pages on mcpmux.com return 404 for entries merged to main after ~2026-02-21, even though the registry files are present and CI passes. Search still surfaces these servers, but clicking through lands on a 404.
Affected entries
Scanned every file in servers/ and hit https://mcpmux.com/servers/<id>. Only two real entries 404:
Every server merged on/before 2026-02-21 resolves with HTTP 200.
Evidence that the site build is stale
The HTML of working detail pages embeds an updated_at timestamp. The newest value I can find across working pages is 2026-02-21T00:56:42Z — matching the last batch of server additions before the gap. Spot-check:
com.atlassian-mcp → updated_at: 2026-02-08T16:00:01.271Zcom.clerk-mcp → updated_at: 2026-02-21T00:55:35.214Zcakerepository.1password-mcp-npx → updated_at: 2026-02-21T00:56:42.228Z
Detail pages appear to be statically generated (Next.js), so servers added after the last build have no pre-rendered page — hence 404. Search hits a fresher data source (presumably the R2-hosted bundle), which is why affected servers still appear in results.
Likely cause
The bundle-build / site-deploy pipeline that's supposed to run on merges to main hasn't produced a new deploy since late February, despite multiple PRs merging since then. This repo's only workflow is validate-pr.yml, so the deploy must live elsewhere (private repo or R2 automation) — worth checking whether it's failing silently or not wired to trigger.
Ask
- Trigger a fresh build/deploy of mcpmux.com so the two pending entries render
- Check why the auto-deploy hasn't been firing on merges to
main for ~7+ weeks
Happy to provide anything else useful.
Summary
Server detail pages on mcpmux.com return 404 for entries merged to
mainafter ~2026-02-21, even though the registry files are present and CI passes. Search still surfaces these servers, but clicking through lands on a 404.Affected entries
Scanned every file in
servers/and hithttps://mcpmux.com/servers/<id>. Only two real entries 404:com.instantdomainsearch-mcp-httpio.wopee-mcpEvery server merged on/before 2026-02-21 resolves with HTTP 200.
Evidence that the site build is stale
The HTML of working detail pages embeds an
updated_attimestamp. The newest value I can find across working pages is 2026-02-21T00:56:42Z — matching the last batch of server additions before the gap. Spot-check:com.atlassian-mcp→updated_at: 2026-02-08T16:00:01.271Zcom.clerk-mcp→updated_at: 2026-02-21T00:55:35.214Zcakerepository.1password-mcp-npx→updated_at: 2026-02-21T00:56:42.228ZDetail pages appear to be statically generated (Next.js), so servers added after the last build have no pre-rendered page — hence 404. Search hits a fresher data source (presumably the R2-hosted bundle), which is why affected servers still appear in results.
Likely cause
The bundle-build / site-deploy pipeline that's supposed to run on merges to
mainhasn't produced a new deploy since late February, despite multiple PRs merging since then. This repo's only workflow isvalidate-pr.yml, so the deploy must live elsewhere (private repo or R2 automation) — worth checking whether it's failing silently or not wired to trigger.Ask
mainfor ~7+ weeksHappy to provide anything else useful.