llm-wiki

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The skill prompt instructs the agent to ask the user for their Obsidian account email and password, and then execute a command embedding the plain-text password (ob login --email <email> --password '<password>'). This requires the LLM to handle user passwords directly in its output/commands.

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

• Third-party content exposure detected (low risk: 0.10). The workflow ingests web pages and PDFs via web_extract when processing user-provided sources into the wiki.

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

• Attempt to modify system services in skill instructions detected (high risk: 1.00). The skill prompt instructs the agent to run sudo loginctl enable-linger $USER, which specifically requires sudo privileges to modify system-level user lingering state.