The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill frequently invokes npx goose-aeo@latest, which downloads and executes the latest version of an external package from the NPM registry at runtime. This introduces risks associated with remote code execution and supply chain integrity.
[COMMAND_EXECUTION]: The skill utilizes shell commands to manage its lifecycle, including npx for tool execution, node -e for environment variable validation, and echo for writing configuration and secrets to disk.
[CREDENTIALS_UNSAFE]: The skill prompts the user to provide several high-value API keys (OpenAI, Perplexity, Gemini, Grok, Claude, DeepSeek, Firecrawl) and instructs the agent to store these secrets in a local .env file.
[EXTERNAL_DOWNLOADS]: Fetches and installs the goose-aeo package from an external repository during the setup and analysis phases.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection (Category 8). It scrapes external website content during the 'Website Audit' phase and processes this untrusted data to generate scores and recommendations.
Ingestion points: Scrapes website pages via the audit command.
Boundary markers: None explicitly defined for the scraped content.
Capability inventory: Executes shell commands and writes to the file system.
Sanitization: No evidence of sanitization or filtering of the scraped content before processing.

aeo