ai-do

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

• Third-party content exposure detected (high risk: 0.90). The skill's SKILL.md (Step 2.5 "Read the FULL candidate skill before writing any prompt" / "How to read: local first, then GitHub") explicitly requires the agent to fetch and read public skill files from GitHub raw URLs (public, user-authored content) before generating prompts, meaning untrusted third‑party content is ingested and can directly influence routing decisions and prompt construction.

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

• Potentially malicious external URL detected (high risk: 0.90). The SKILL.md explicitly instructs fetching remote skill files at runtime (e.g., https://raw.githubusercontent.com/lebsral/DSPy-Programming-not-prompting-LMs-skills/main/skills//SKILL.md) and using that fetched content to craft the agent's prompts, so this external raw.githubusercontent.com URL is a required runtime dependency that directly controls prompts.