skill-best-practices
Fail
Audited by Gen Agent Trust Hub on May 11, 2026
Risk Level: HIGHCOMMAND_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: In
SKILL.md, the instruction for Step 1.3 directs the agent to execute a shell command:python3 scripts/validate-metadata.py --name "[name]" --description "[description]". This pattern is susceptible to command injection because the placeholders[name]and[description]are directly interpolated into the command string. If an attacker provides metadata containing shell metacharacters (such as;,&,|, or backticks), they can break out of the intended command and execute arbitrary code on the host system. - [COMMAND_EXECUTION]: The template provided in
assets/SKILL.template.mdreinforces dangerous patterns by instructing the agent to execute local scripts via shell commands (e.g.,Execute python scripts/[script-name].py). Without explicit instructions to sanitize user input before passing it to these scripts, this promotes a workflow that is inherently vulnerable to exploitation through malicious data ingestion.
Recommendations
- AI detected serious security threats
Audit Metadata