paper-release-checklist
Pass
Audited by Gen Agent Trust Hub on Sep 9, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill features a feedback loop where user-provided descriptions of new mistakes are appended to the
references/checklist.mdfile. This file is used as a reference for the agent's checklist, creating a risk that a malicious user could store instructions that the agent might later follow during subsequent interactions. - Ingestion points: User-reported mistakes that are not currently in the checklist (SKILL.md, Step 5).
- Boundary markers: None specified; user input is recorded verbatim into the reference document.
- Capability inventory: Access to append data to the file system (
references/checklist.md) and command execution capabilities (latex-release-lint). - Sanitization: No sanitization, validation, or escaping of the user-provided content is mentioned before it is written to the file.
- [COMMAND_EXECUTION]: The skill instructs the agent to execute a command-line utility called
latex-release-lint. This tool is used to automate mechanical checks on LaTeX sources, representing a standard use of external tools for linting tasks.
Audit Metadata