The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill instructs the user or agent to install the skit CLI by downloading a script from a remote URL and piping it directly into the shell. This pattern executes unverified code from a remote source at runtime.
Evidence: curl -fsSL https://raw.githubusercontent.com/vlln/skit/main/install.sh | sh in SKILL.md.
[COMMAND_EXECUTION]: The core functionality of the skill relies on executing various shell commands to search for, install, and update other agent skills.
Evidence: Usage of commands such as skit search, skit install, skit list, and skit check.
[EXTERNAL_DOWNLOADS]: The skill fetches external content from GitHub, both for the initial tool installation and when installing specific skills from remote repositories.
Evidence: URL https://raw.githubusercontent.com/vlln/skit/main/install.sh and variable source selectors like <source@skill>.
[PROMPT_INJECTION]: The skill has an indirect prompt injection surface because it ingests untrusted data from the internet via search results and processes it to make installation recommendations.
Ingestion points: Output from the skit search <query> command.
Boundary markers: Absent; the agent is not instructed to treat search output as untrusted or to ignore embedded instructions.
Capability inventory: The agent has the ability to install and execute additional code (skills) and modify the local environment via the skit tool.
Sanitization: No sanitization or validation of the search results is specified before the agent presents options to the user.

search-skills