ai-do

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

• Third-party content exposure detected (high risk: 0.90). SKILL.md explicitly instructs the agent to fetch and read SKILL.md and other supporting files from public GitHub URLs (https://github.com/... and https://raw.githubusercontent.com/...), which are user-authored, untrusted third-party pages that the agent must interpret to choose skills and craft runnable prompts, enabling indirect prompt injection.

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

• Potentially malicious external URL detected (high risk: 0.80). The skill explicitly instructs fetching skill files from GitHub at runtime (e.g. https://raw.githubusercontent.com/lebsral/DSPy-Programming-not-prompting-LMs-skills/main/skills//SKILL.md and the directory listing URL https://github.com/lebsral/DSPy-Programming-not-prompting-LMs-skills/tree/main/skills/) and those remote files are read to craft the agent's prompts, so external content can directly control prompts.