muapi-social-media-video
Pass
Audited by Gen Agent Trust Hub on Oct 1, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests content from external markdown files (brand-identity.md, ICP.md, messaging.md) and uses this data to construct prompts for subsequent image and video generation steps. This creates a surface for indirect prompt injection attacks where malicious instructions embedded in these files could influence agent behavior.
- Ingestion points: Brand identity, ICP, and messaging files are read during Step 1 of the execution protocol in SKILL.md.
- Boundary markers: No explicit delimiters or instructions are provided to the agent to treat the content of these files as untrusted data or to ignore embedded commands.
- Capability inventory: The skill executes shell scripts (generate-image.sh, run-social-video.sh) that perform file system writes and network operations via curl.
- Sanitization: No sanitization or validation of the input file content is performed before interpolation into prompts.
- [EXTERNAL_DOWNLOADS]: The orchestration script performs automated downloads of generated media assets from dynamic URLs.
- Evidence: The run-social-video.sh script uses curl to download images returned by the generation pipeline.
Audit Metadata