beatra-ai-video-studio
Warn
Audited by Gen Agent Trust Hub on Nov 14, 2026
Risk Level: MEDIUMEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONDATA_EXFILTRATIONCREDENTIALS_UNSAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill directs the agent to clone a remote repository from GitHub (
https://github.com/beatra-ai/beatra-skills.git) during the installation phase. - [REMOTE_CODE_EXECUTION]: The installation flow requires executing downloaded Python scripts (
authorize.py,mcp_client.py,uninstall.py) on the host machine. - [DATA_EXFILTRATION]: The skill collects and transmits telemetry data to external domains (
api.beatra.aiandapi.beatra.cn), including the system hostname, OS version, CPU architecture, and locale information. - [CREDENTIALS_UNSAFE]: The skill manages sensitive API bearer tokens by storing them in a local file at
~/.beatra/credentials.json. - [INDIRECT_PROMPT_INJECTION]: The skill acts as an ingestion surface for untrusted data by processing user prompts and media files for transmission to third-party model providers.
- Ingestion points: User prompts and media files processed for video generation.
- Boundary markers: None specified in the SKILL.md instructions.
- Capability inventory: Python scripts maintain network capabilities (HTTPS) and file system access for local state management in
~/.beatra. - Sanitization: Not documented in the provided skill file.
- [COMMAND_EXECUTION]: The instructions utilize several shell commands for environment setup, file inspection, and script execution, including
git,python3,grep, andawk.
Audit Metadata