performing-paste-site-monitoring-for-credentials

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The skill's code extracts and records credential "samples" from pastes and includes them in alerts/JSON output (and uses an explicit GitHub token header), which requires handling and outputting secret values verbatim.

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

• Third-party content exposure detected (high risk: 1.00). The skill explicitly fetches and analyzes public, user-generated content (e.g., Pastebin via SCRAPING_URL in SKILL.md's PastebinMonitor.fetch_recent_pastes/get_paste_content, HaveIBeenPwned paste API in scripts/agent.py.check_paste_dumps, and GitHub gists via GitHubSecretMonitor.search_gists), and those untrusted pastes/gists are parsed to drive severity scoring and generate alerts/actions, so third-party content can materially influence agent behavior.