to-seams

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 0.90). The skill requires the agent to accept user-pasted interface text "verbatim" and to display/write that exact content (the seam file and appended [[seam]] block), so any secrets included in that pasted text would be output and stored verbatim, creating an exfiltration risk.

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (medium risk: 0.30). The skill records a seam by asking the user to paste verbatim interface text (consumer-prompt ground truth) and then writes it to <workspace>/.claude/seams/<name>, meaning outsider-authored free text can be ingested into an agent’s runtime prompt later.