eval-config
Fail
Audited by Gen Agent Trust Hub on Oct 13, 2026
Risk Level: HIGHCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill constructs and executes shell commands that interpolate several user-controlled variables, including
{slug},{dimension}, and{weights_json}. Specifically, in theset-weightsprocess, it executespython "${CLAUDE_PLUGIN_ROOT}/scripts/eval-config-manager.py" --brand {slug} --action set-weights --weights '{weights_json}'. If these variables are not strictly validated or escaped, an attacker could provide input containing shell metacharacters to execute unauthorized commands. - [INDIRECT_PROMPT_INJECTION]: The skill processes data from local files such as
profile.json,_manifest.json, and agency SOPs in~/.claude-marketing/sops/. These files constitute an attack surface where maliciously crafted content could influence the agent's scoring recommendations or configuration logic.
Recommendations
- AI detected serious security threats
Audit Metadata