anthropic-skilljar-extractor

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

• Third-party content exposure detected (high risk: 0.90). The skill's required workflow and script (SKILL.md and scripts/extract_course.py) fetch and parse HTML from anthropic.skilljar.com (via fetch()) and extract window.__chatData and CloudFront-hosted images (extract_chat_data, extract_section_images), then interpret those untrusted third-party page contents to build lesson markdown—so external page content can directly influence parsing, matching, and subsequent actions.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.90). The script fetches the user-provided Anthropic Skilljar course URL (https://anthropic.skilljar.com/...) at runtime and extracts window.__chatData (embedded LLM notes) which are parsed and injected into the generated lesson markdown, meaning remote page content can directly control the skill's outputs.