The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill provides instructions for the user or agent to install the inference.sh CLI tool by piping a remote script from an external URL directly into the shell (curl -fsSL https://cli.inference.sh | sh). This execution pattern allows the remote server to run arbitrary code on the local system without verification.
[COMMAND_EXECUTION]: The skill makes extensive use of the Bash tool to run infsh commands. It includes instructions for complex shell workflows, such as utilizing for loops and output redirection to write JSON files to the local file system.
[EXTERNAL_DOWNLOADS]: The skill references and downloads binaries, checksums, and media assets from several external domains, including cli.inference.sh, dist.inference.sh, cloud.inference.sh, and third-party media URLs (e.g., trending-music.mp3).
[INDIRECT_PROMPT_INJECTION]: The skill possesses a vulnerability surface for indirect prompt injection due to the interpolation of untrusted data into shell commands. 1. Ingestion points: User-supplied prompts for video generation and shell loop variables ($section) are interpolated into infsh command arguments. 2. Boundary markers: No boundary markers or delimiters are used to isolate untrusted data. 3. Capability inventory: The skill uses Bash(infsh *) to perform generative tasks, file operations, and network requests. 4. Sanitization: No input validation or escaping mechanisms are implemented for the interpolated prompts.

ai-marketing-videos