cite-check
Pass
Audited by Gen Agent Trust Hub on Sep 7, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied markdown and bib files, interpolating content into LLM prompts.\n
- Ingestion points: Reads files from the drafts directory and bibliography entries from provided .bib files.\n
- Boundary markers: None identified; claim and context text are directly embedded in the verification prompt.\n
- Capability inventory: Shell command execution (rclone, readwise, python3), file writing, and network API access.\n
- Sanitization: The skill does not sanitize or escape extracted text before interpolation.\n- [COMMAND_EXECUTION]: The skill executes external CLI tools (rclone, readwise) using
spawnSyncwith arguments such as titles and file paths from bib files, which could be exploited via argument injection if the bib files are malicious.\n- [EXTERNAL_DOWNLOADS]: The skill relies on host-installed tools likercloneand thereadwiseCLI, and thepymupdf4llmPython package. It performs network downloads from Google Drive and Readwise APIs to synchronize source documents.
Audit Metadata