fit-scorer
Pass
Audited by Gen Agent Trust Hub on Sep 2, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill utilizes a local Python script to fetch influencer data, passing user-provided handles as command-line arguments.
- Evidence:
python3 "${CLAUDE_PLUGIN_ROOT}/scripts/connectors/youtube.py" videos @handle --limit 10inSKILL.md. - Security Concern: User-supplied input (
@handle) is interpolated directly into a shell command. While intended for legitimate data fetching, this represents a command injection surface if the agent does not strictly validate the handle format before execution. - [INDIRECT_PROMPT_INJECTION]: The skill is designed to process external, untrusted data from social media platforms to perform suitability scoring.
- Ingestion points: The skill reads influencer handles, social platform analytics (e.g., "comment quality samples"), and audience intelligence from external sources.
- Boundary markers: No explicit delimiters or instructions to ignore embedded commands are present in the provided templates to protect the agent from malicious content within social media comments or profiles.
- Capability inventory: The skill has the ability to execute shell commands (via the YouTube connector) and write files to the local memory filesystem (
memory/influencer/fit-scorer/). - Sanitization: There is no evidence of sanitization, escaping, or validation logic to filter potentially malicious instructions embedded in the external data being analyzed.
Audit Metadata