Skills
Navigate
skills/noe-varner/noeai-free-resources/clipping/Gen Agent Trust Hub

clipping

Warn

Audited by Gen Agent Trust Hub on Jul 30, 2026

Risk Level: MEDIUMCOMMAND_EXECUTIONDATA_EXFILTRATIONCREDENTIALS_UNSAFEREMOTE_CODE_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [COMMAND_EXECUTION]: Instructions in SKILL.md and references/testimonials.md direct the agent to execute ffmpeg with several variable-based arguments and local Python scripts.
  • [DATA_EXFILTRATION]: The references/testimonials.md file contains a shell script that uses curl to upload local binary video data to an external Dropbox endpoint.
  • [CREDENTIALS_UNSAFE]: The skill requires access to and usage of high-privilege credentials like SUPABASE_SERVICE_KEY and Dropbox application secrets stored in .env files.
  • [REMOTE_CODE_EXECUTION]: A complex shell command pipeline in references/testimonials.md executes a dynamic Python one-liner (python3 -c) to parse responses from a remote API.
  • [EXTERNAL_DOWNLOADS]: The Python scripts (scripts/vault_embed.py and scripts/vault_search.py) make network requests to external domains (api.openai.com and Supabase URLs) to perform embeddings and searches.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Jul 30, 2026, 03:50 PM