The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill contains instructions to download additional components from the author's GitHub repository (https://github.com/ulpi-io/skills). Since these resources belong to the vendor 'ulpi-io', they are considered part of the intended product ecosystem.
[REMOTE_CODE_EXECUTION]: The instructions feature the command npx skills add pointing to a remote repository URL. This facilitates the execution of external code to provide companion functionality.
[COMMAND_EXECUTION]: The skill describes using the npx CLI to install dependencies, which involves running shell commands with remote arguments.
[PROMPT_INJECTION]: This skill presents an indirect prompt injection risk by ingesting untrusted data from live websites and local repository files without sanitization. Ingestion points: Step 1 involves browsing the product's live site and reading repository files like the README. Boundary markers: No explicit delimiters or instructions are used to separate untrusted data from the system prompt. Capability inventory: The skill utilizes the Write and Skill tools to perform actions based on processed data. Sanitization: There is no verification or escaping of external content before it is used to generate the launch package.

launch-linkedin