bioinformatics

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

• Third-party content exposure detected (high risk: 0.85). The skill is a “gateway” that clones and then reads other repos’ SKILL.md/README at runtime; those fetched files are outsider-authored (GPTomics/bioSkills and ClawBio/ClawBio are not the operating user’s own content) and are ingested as readable text into the agent’s LLM context via the “cat /tmp/.../SKILL.md” / “cat /tmp/.../README.md” step.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill instructs cloning and using external GitHub repositories at runtime (https://github.com/GPTomics/bioSkills and https://github.com/ClawBio/ClawBio), and ClawBio explicitly contains runnable scripts/pipelines (and recommends installing and running them), so fetching those repos will pull remote code that can be executed and directly control runtime behavior.

MEDIUM W013: Attempt to modify system services in skill instructions.

• Attempt to modify system services in skill instructions detected (medium risk: 0.60). The skill instructs the user to perform system-level package installs (e.g., "sudo apt install", brew/conda installs) and other environment-changing commands which require elevated privileges and modify the host system state, so it encourages actions that can compromise the machine.