skills/yrzhe/claude-skills/seed/Gen Agent Trust Hub

seed

Pass

Audited by Gen Agent Trust Hub on Jun 27, 2026

Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses subprocess.run to execute local scripts and system utilities for its primary functions.
  • The capture-session-seed.sh hook executes scripts/extract_turn.py to process and store session transcripts.
  • The scripts/shot.py script executes the macOS screencapture utility for interactive window captures.
  • The scripts/shot.py script executes local browser binaries (Google Chrome, Microsoft Edge, etc.) in headless mode to capture screenshots of provided URLs.
  • [DATA_EXPOSURE]: The skill maintains a local repository of all Claude Code session history (user prompts, tool usage, and assistant responses) in ~/.claude/skills/seed/state/sessions/. While this is the intended purpose for building-in-public activity logging, users should be aware that these local files contain their conversation history.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 27, 2026, 07:09 AM
Security Audit — agent-trust-hub — seed