unfreeze

Pass

Audited by Gen Agent Trust Hub on Jun 14, 2026

Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
  • [COMMAND_EXECUTION]: Executes shell commands to create local directories, log usage metadata to a JSONL file in the user's home directory, and delete internal state files.
  • [DYNAMIC_EXECUTION]: Uses eval to source environment variables from a local script located at ~/.claude/skills/gstack/bin/gstack-paths. This is a standard pattern for CLI tool configuration.
  • [INDIRECT_PROMPT_INJECTION]:
  • Ingestion points: Reads directory paths from the local file $STATE_DIR/freeze-dir.txt and echoes them.
  • Boundary markers: Absent.
  • Capability inventory: The skill uses Bash and file system tools (Read, Write, Edit).
  • Sanitization: None. However, as the source is a local system state file, the risk of injection is negligible.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 14, 2026, 08:50 AM
Security Audit — agent-trust-hub — unfreeze