#539 Lab profanity filter rejects legitimate CLI flag tokens via substring match

Steps to reproduce

1. swamp auth login
2. Submit a feature or bug report whose body contains a literal long-form CLI flag for Apple's spctl(8) Gatekeeper verification subcommand — two hyphens followed by the English verb meaning "to evaluate" (spelled with the same three-letter substring that the filter treats as profanity; the substring is the common slang term for the rear of an animal).

Expected behavior

Submission succeeds. The flag is a legitimate UNIX option documented in Apple's man pages, and the underlying English word is neutral.

Actual behavior

Server returns HTTP 422:

{"error": "Description contains inappropriate language", "flagged": ["--<the-flag>"]}

The filter appears to substring-match the three-character root without word-boundary checks. Many common neutral English words also contain this substring (verbs meaning "to help" or "to claim", nouns meaning "a moving-through" or "a category" or "a quantity"), so legitimate prose using any of those words would also be rejected.

This makes the Lab unusable for filing issues that legitimately need to reference those CLI flags or words — including, for example, anything about macOS code signing or Gatekeeper verification.

Environment

• swamp version: 20260527.235818.0-sha.ed572875
• platform: macOS

Suggested fix

Apply word-boundary anchors to the profanity check, or switch to a tokenized check rather than naive substring search. Failing that, allowlist common English words containing the offending substring.