pmem-style-generation

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

• Third-party content exposure detected (high risk: 0.80). The skill explicitly fetches and ingests arbitrary webpage content when a user provides a URL (SKILL.md step "If user provides a URL, first call baoyu-url-to-markdown" with the npx command to save /tmp/source_content.md and the "从 URL 生成" example saying "获取网页内容"), so untrusted third‑party pages are read and their content is used in the extraction+generation workflow and can therefore influence subsequent prompt assembly and actions.

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

• Potentially malicious external URL detected (high risk: 1.00). The skill runs a runtime fetch of arbitrary user-provided webpages via the command "npx -y bun ${SKILL_URL_DIR}/scripts/main.ts -o /tmp/source_content.md" (example shown as https://example.com/article), and the fetched page content is directly injected as {MATERIAL} into the model prompts for extraction and generation, which can be used for prompt injection or to control model outputs.