wiki

Fail

Audited by Gen Agent Trust Hub on Sep 26, 2026

Risk Level: HIGHEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPRIVILEGE_ESCALATIONPERSISTENCEDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [PRIVILEGE_ESCALATION]: The skill provides instructions to execute administrative commands using sudo (specifically sudo loginctl enable-linger $USER) to modify session behavior and allow background processes to persist.
  • [PERSISTENCE]: Includes configuration for a systemd user service (~/.config/systemd/user/obsidian-wiki-sync.service) to maintain background synchronization tasks across user sessions.
  • [COMMAND_EXECUTION]: Instructions direct the installation of external global packages via npm and the execution of various CLI tools for synchronization and login purposes.
  • [EXTERNAL_DOWNLOADS]: Fetches and installs third-party software (obsidian-headless) from the NPM registry to enable headless synchronization features.
  • [DYNAMIC_EXECUTION]: The skill utilizes the execute_code tool to run Python scripts for wiki maintenance tasks, such as link validation, orphan page detection, and knowledge graph analysis.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from external sources.
  • Ingestion points: External URLs are processed via web_extract and stored as raw markdown files.
  • Boundary markers: While YAML frontmatter is used, there are no explicit boundary markers or instructions to disregard potentially malicious instructions within the ingested source material.
  • Capability inventory: The agent has access to execute_code, write_file, and web_extract tools.
  • Sanitization: No sanitization or filtering of the raw external content is performed before the agent reads and synthesizes the information into the wiki.
Recommendations
  • AI detected serious security threats
Audit Metadata
Risk Level
HIGH
Analyzed
Sep 26, 2026, 11:55 AM