The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill makes extensive use of npx -y @steipete/oracle, which downloads an external package from the NPM registry and executes it immediately without user confirmation of the package source.
[REMOTE_CODE_EXECUTION]: By using the npx command with external packages, the skill performs remote code execution of third-party code on the local system.
[COMMAND_EXECUTION]: The instructions direct the agent to execute shell commands to bundle files and interact with external processes (browsers or APIs).
[DATA_EXFILTRATION]: The core functionality of the skill involves reading local source files (e.g., src/**) and transmitting their content to external AI platforms such as OpenAI's ChatGPT or other model APIs.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection (Category 8) because it ingests untrusted data from the local file system and interpolates it into a prompt for a secondary AI model.
Ingestion points: Local files matched by the --file patterns in SKILL.md.
Boundary markers: None explicitly defined in the CLI arguments provided.
Capability inventory: The @steipete/oracle tool has capabilities for file system access, network communication, and browser automation.
Sanitization: There is no evidence of sanitization or filtering applied to the contents of the bundled files before they are sent to the target AI model.

oracle