summarize

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

• Third-party content exposure detected (high risk: 0.90). The skill explicitly accepts and converts arbitrary public URLs via uvx markitdown (see SKILL.md and to-markdown.mjs runMarkitdown) and then feeds the resulting untrusted Markdown into the summarizer (summarizeWithPi) which the model reads/interprets to produce summaries, allowing third-party page content to influence the agent's outputs.

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

• Potentially malicious external URL detected (high risk: 1.00). The skill runs "uvx markitdown" on a user-supplied https:// URL at runtime (e.g., https://example.com) and then injects the fetched Markdown directly into the prompt sent to the "pi" CLI, so remote content can directly control model instructions.