The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill recommends the installation and execution of third-party tools from unverified sources. Specifically, it suggests using npx gooseworks install --all in references/imessage-video-ads.md and cloning/building a repository from an unverified GitHub account (git clone https://github.com/jamiepine/voicebox.git) in references/generative-tools.md.
[EXTERNAL_DOWNLOADS]: The skill facilitates downloading software and assets from various external sources not listed as trusted vendors, including personal GitHub repositories and third-party media sites like bigsoundbank.com.
[COMMAND_EXECUTION]: The skill instructs the agent to run various shell commands, including local scripts for interacting with advertising APIs (e.g., node tools/clis/google-ads.js) and build tools for external repositories.
[INDIRECT_PROMPT_INJECTION]: The skill design relies on ingesting external, untrusted data such as customer reviews and ad comments from files in the inputs/ directory. 1. Ingestion points: Files in inputs/reviews/ and inputs/comments/ as described in SKILL.md. 2. Boundary markers: No specific boundary markers or instructions to ignore embedded commands are provided for these ingestion points. 3. Capability inventory: The agent has the ability to execute platform-specific CLI tools, write files to the local system, and generate HTML review pages. 4. Sanitization: The skill lacks explicit instructions for sanitizing or validating the content of ingested reviews and comments before processing.

ad-creative