etetoolkit

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

• Third-party content exposure detected (medium risk: 0.30). The skill’s runtime workflow reads outsider-provided Newick alignment/tree text from user-supplied files/strings (e.g., scripts/quick_visualize.py calls Tree(str(tree_file), ...) on input, and scripts/tree_operations.py loads the input tree file and can also read taxa lists from a provided file path), which is ingested directly by the LLM as free text contained in those files.