Skip to content

New rule: detect suspicious encoded blobs in comments (beyond base64) #19

Open

Description

@sarahxsanders

Context

From Vincent's review on PR #12 — the current prompt_injection_base64_in_comment rule only catches base64. Other encoding schemes could also be used to smuggle payloads in comments.

What to catch

  • Long hex strings in comments (e.g., \x48\x65\x6c\x6c\x6f)
  • URL-encoded blobs (%48%65%6C%6C%6F)
  • Other obfuscation patterns (rot13, unicode escapes, etc.)

Considerations

  • High false-positive risk. Legitimate code comments often contain:
    • SHA hashes (40+ hex chars)
    • JWTs in test fixtures
    • UUIDs
    • Color codes, memory addresses
  • Needs careful threshold tuning per encoding type.
  • Better as companion rules (one per encoding type) rather than broadening the existing base64 rule.

Origin

PR #12 review comment by @gewenyu99

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    enhancementNew feature or request

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions