skill-linker

Warn

Audited by Gen Agent Trust Hub on Jun 19, 2026

Risk Level: MEDIUMCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The script scripts/install.sh utilizes sudo to invoke system package managers (apt-get, dnf, pacman) for installing the fzf utility if it is not already present on the system.
  • [COMMAND_EXECUTION]: The installation script modifies several user shell profile files, including .bashrc, .zshrc, and .bash_profile, to export the SKILL_LINK_ROOT environment variable and append ~/.local/bin to the PATH.
  • [COMMAND_EXECUTION]: The core utility scripts/skill-link.sh performs file system manipulations by creating and removing symbolic links within configuration directories for editors such as Cursor, Claude Code, and Codex.
  • [PROMPT_INJECTION]: The skill presents an indirect prompt injection surface by reading and displaying arbitrary file names from a user-defined directory to the agent.
  • Ingestion points: The collect_items function in scripts/skill-link.sh retrieves a list of directory and file names from the path stored in SKILL_LINK_ROOT.
  • Boundary markers: No specific delimiters or security warnings are provided in the instructions to prevent the agent from being influenced by malicious file names.
  • Capability inventory: The skill possesses file system modification capabilities (creating symbolic links) and can trigger script installations.
  • Sanitization: There is no validation or filtering performed on the file names retrieved from the source directory before they are displayed or used in link operations.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Jun 19, 2026, 10:56 AM
Security Audit — agent-trust-hub — skill-linker