paperzilla
Pass
Audited by Gen Agent Trust Hub on Sep 25, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONREMOTE_CODE_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill provides instructions to install the 'pz' CLI tool from third-party repositories on GitHub under the 'paperzilla-ai' organization using Homebrew or Scoop.
- [COMMAND_EXECUTION]: The skill provides commands for shell execution, including authentication ('pz login') and administrative updates ('pz update').
- [REMOTE_CODE_EXECUTION]: The 'pz update' command facilitates the download and execution of updated binary versions from remote servers.
- [INDIRECT_PROMPT_INJECTION]: The skill reads research papers and project feeds via the CLI to generate summaries and analysis, providing a surface for malicious instructions embedded in external research data.
- Ingestion points: Outputs from 'pz feed', 'pz paper', and 'pz rec' commands.
- Boundary markers: None specified in instructions.
- Capability inventory: Shell command execution, file system interaction via the CLI tool.
- Sanitization: No evidence of validation or sanitization of the markdown or JSON data retrieved from the Paperzilla API.
Audit Metadata