I built a checker and ran the 50 most-copied templates on n8n.io. 10 of 50 won’t import-and-run unmodified, and a few run green while silently producing wrong output. Sharing the three biggest, since a lot of people here probably use them.
Same root cause in all three: an expression field missing its leading =, so n8n stores the {{ }} as literal text instead of evaluating it.
Angie, personal AI assistant (368k views): the agent’s systemMessage ships Today's date is {{ $now }} with no =, so the assistant is told its own date is the literal string {{ $now }}. Anything date-aware is quietly wrong. Fix: put a leading = on the field so the interpolation runs.
Generate AI videos with Veo3, save to Drive, upload to YouTube (271k views): the Update YouTube URL step writes https://youtu.be/{{ $json.results.youtube.video_id }} to the sheet verbatim, braces and all, every run. Fix: one character, the leading =.
Automated job-applications tracker (95k views): seven fields unevaluated, including email subjects sent as Status Update: {{ $json.Company }} - {{ $json.Position }} literally, and Sheets ranges like D{{ $json.row_index }}:F{{ ... }}. Six fix cleanly with the =; the seventh (a cover-letter field) uses bare {{position}}/{{company}} and also needs the right $json reference.
The fix is basically one character each. The point is that nothing in n8n warns you: it imports, every node is green, it “works,” and it’s wrong.
PSA: if you run a popular template, check whether your expression fields actually start with =. Happy to share the fixed JSONs if anyone wants them.