context-resume

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The skill explicitly instructs the agent to read the selected session file and "完整展示内容" (display the session file contents verbatim) and to update session files, which forces the LLM to output any secret values present in those files directly.

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

• Third-party content exposure detected (medium risk: 0.65). The required workflow reads and fully displays Markdown session files from the docs/context-sessions/ directory (Step 2), and those files could be outsider-authored/modified free-text that the agent then uses to guide subsequent actions—an indirect prompt-injection vector.