The public generateCurl helper uses -d, so an @-prefixed body is read as a local filename instead of sent literally. It drops empty bodies, silently truncates long bodies, and a GET with a body becomes POST because -X GET is omitted. A URL beginning with a dash is also interpreted as an option. I plan to use --data-raw, retain explicit methods when a body is present, preserve complete bodies in the standalone helper, and use --url. Inspector capture limits will remain a separate, clearly labelled concern. Validation will inspect shell-parsed argv without contacting any service.
The public generateCurl helper uses -d, so an @-prefixed body is read as a local filename instead of sent literally. It drops empty bodies, silently truncates long bodies, and a GET with a body becomes POST because -X GET is omitted. A URL beginning with a dash is also interpreted as an option. I plan to use --data-raw, retain explicit methods when a body is present, preserve complete bodies in the standalone helper, and use --url. Inspector capture limits will remain a separate, clearly labelled concern. Validation will inspect shell-parsed argv without contacting any service.