byted-sol-stability-architecture-path-extractor

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

• Third-party content exposure detected (high risk: 0.75). Yes—outsider-authored free text from the user-supplied repository/docs/diagram files (e.g., product-docs and arch-diagrams paths) is read as UTF-8 prose at runtime (e.g., parse_doc_diagram_signals → p.read_text(...) and parse_api_signals/parse_config_signals → f.read_text(...)), then its extracted lines/keywords are carried into the model outputs as evidence that would be included in the LLM context if the agent uses those outputs for prompting.