lov-media-crawler

Warn

Audited by Gen Agent Trust Hub on Aug 24, 2026

Risk Level: MEDIUMEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill downloads the MediaCrawler repository from GitHub (https://github.com/NanmiCoder/MediaCrawler.git) during the setup process. This is an external, unverified dependency.\n- [REMOTE_CODE_EXECUTION]: The skill executes code from the downloaded MediaCrawler repository using uv run with a dynamically generated command string.\n- [COMMAND_EXECUTION]: Several scripts execute shell commands through subprocesses, including git, uv, curl, aria2c, ffprobe, and the macOS security tool.\n- [DATA_EXFILTRATION]: The skill allows sending social media URLs to a third-party worker service (sph.litao.workers.dev) for resolution, which is gated by an optional flag but remains a potential exfiltration vector for user activity metadata.\n- [PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection by processing untrusted metadata from social platforms.\n
  • Ingestion points: Metadata such as titles and descriptions are fetched from external platforms in scripts/media_crawler.py.\n
  • Boundary markers: The skill does not use specific delimiters or instructions to prevent the agent from interpreting ingested metadata as commands.\n
  • Capability inventory: The skill can write files, make network requests, and execute subprocesses across all scripts.\n
  • Sanitization: External metadata is used directly without sanitization, creating a surface for prompt injection if the agent processes the resulting report.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Aug 24, 2026, 12:55 AM
Security Audit — agent-trust-hub — lov-media-crawler