publish-and-analyze
Pass
Audited by Gen Agent Trust Hub on Oct 10, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The
scripts/dupe_check.pyscript usessubprocess.runto callffmpegandffprobefor media processing. The implementation uses list-based argument passing rather than a shell string, which is a secure practice to prevent shell injection attacks. - [EXTERNAL_DOWNLOADS]: The documentation in
references/favstash-connection.mdguides users to install the@sketric/favstash-mcppackage vianpmto enable scheduling and analytics features. This external dependency is appropriate for the skill's stated purpose of integrating with the FavStash service. - [INDIRECT_PROMPT_INJECTION]: The
dupe_check.pyscript serves as a data ingestion point because it processes external media URLs and manifest files. While this represents a theoretical surface for indirect prompt injection (e.g., via malicious media metadata), the skill effectively mitigates this risk by enforcing a human-in-the-loop approval process before any publishing actions occur. - Ingestion points: The
scripts/dupe_check.pyscript accepts JSON manifests and remote HTTPS URLs for video hashing. - Boundary markers: Instructions in
SKILL.mdexplicitly state that nothing is uploaded or scheduled until the creator approves the exact package and settings. - Capability inventory: The skill performs local file system operations (caching fingerprints in
~/.cache/dupe-check) and executes media processing commands viasubprocess. - Sanitization: The Python script validates inputs by checking for file existence and using secure subprocess call patterns to avoid command injection.
Audit Metadata