latex-conference-template-organizer
Pass
Audited by Gen Agent Trust Hub on Mar 8, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: Shell commands are used to manipulate user-supplied files. The loop
while read f; do cp "$f" output/tables/; donelacks robust handling for filenames with special characters or newlines, potentially leading to unintended command behavior or errors if an attacker crafts a malicious archive. - [PROMPT_INJECTION]: The skill processes untrusted external data, creating an attack surface for indirect prompt injection.
- Ingestion points: Processes content from user-uploaded
.ziparchives and fetches text from external URLs provided by the user (conference submission pages). - Boundary markers: No specific delimiters or "ignore instructions" directives are identified to prevent the agent from following instructions embedded within extracted LaTeX files or fetched web content.
- Capability inventory: Includes file system access (read, write, delete), directory creation, archive extraction via
unzip, and shell-based search operations viagrepandfind. - Sanitization: No evidence of sanitization is provided for filenames or the content extracted from the untrusted archives or remote URLs before they are processed by the agent.
Audit Metadata