The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The scripts/install.sh script downloads a pre-compiled binary from https://github.com/gykim80/perfectpixel-studio/releases. This source is an unverified third-party account and not an official or trusted organization.
[REMOTE_CODE_EXECUTION]: The skill instructs the agent to execute the downloaded ppgen binary. Running unverified binaries from untrusted sources can lead to unauthorized code execution and full system compromise.
[COMMAND_EXECUTION]: The install.sh script performs several high-risk commands, including downloading files from the internet using curl, cloning git repositories, and compiling source code using go build.
[DATA_EXFILTRATION]: The skill interacts with sensitive information such as GEMINI_API_KEY, OPENROUTER_API_KEY, and a local authentication file at ~/.codex/auth.json. There is a significant risk that the unverified ppgen binary could exfiltrate these credentials to a remote server.
[PROMPT_INJECTION]: The skill accepts arbitrary user-supplied descriptions via the -desc flag, which are then passed to external AI models. This creates an attack surface for indirect prompt injection as there are no clear boundary markers or sanitization steps mentioned for this input.

perfectpixel