skills/magic5644/skills/skill-manager/Gen Agent Trust Hub

skill-manager

Pass

Audited by Gen Agent Trust Hub on Aug 31, 2026

Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The uninstaller script executes standard shell utilities to manage files in the agent's skill directories.
  • The script uses find to discover installed skills and rm -rf to delete directories selected by the user.
  • These operations are limited to specific skill-related paths and require manual user confirmation before execution.
  • [INDIRECT_PROMPT_INJECTION]: The tool processes directory names from the filesystem which represents a potential attack surface if a malicious skill creates specifically crafted directory names.
  • Ingestion points: Directory names are read from standard skill storage paths (e.g., ~/.copilot/skills) via the find command in scripts/uninstall-interactive.sh.
  • Boundary markers: None; filesystem metadata is processed as raw strings.
  • Capability inventory: The script has the capability to delete files and folders (rm -rf).
  • Sanitization: The risk is mitigated by the use of strict variable quoting in the shell script to prevent command injection via filenames, and the requirement for an interactive user review and confirmation before any destructive action occurs.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 31, 2026, 12:30 PM
Security Audit — agent-trust-hub — skill-manager