remove-ai-tells
Pass
Audited by Gen Agent Trust Hub on Sep 19, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill utilizes common Unix utilities including
bash,grep,awk, andgitto identify text patterns within local documents. These operations are restricted to the local file system and do not request network access or administrative privileges. - [INDIRECT_PROMPT_INJECTION]: The skill scans potentially untrusted document files and outputs the matching text directly into the agent's context. This creates a vulnerability where malicious instructions embedded in a document could be executed by the agent when it processes the scan results.
- Ingestion points: Document files identified via command arguments, IDE state, or git staged files (Step 1).
- Boundary markers: Findings are presented in a structured list, but there are no explicit delimiters or instructions to the agent to ignore command-like text within the matched document snippets.
- Capability inventory: The skill possesses file-reading capabilities (
grep,awk) and file-writing capabilities (Step 5 replacements). - Sanitization: The skill performs no escaping or sanitization of the text found in the target documents before reporting it to the agent.
Audit Metadata