The Agent Skills Directory

[CREDENTIALS_UNSAFE]: The file references/authentication.md contains a hardcoded API key (dot_app_UlSpzXNEXhYZIAFakHLCkMVVBLbsBIWxaRMVaJZGUOYKhDoDRZwLLvLujAIwQxbY) within an example curl command. Instructions specify that keys should be kept secret and stored in environment variables, yet a concrete key is provided in the documentation.
[DATA_EXFILTRATION]: The skill provides scripts such as scripts/send_image.py, scripts/send_text.py, and scripts/send_canvas.py that are designed to read local files (images, icons, or JSON payloads) and transmit their content to the vendor's API at dot.mindreset.tech. While this is functional for the skill's purpose, the capability to read and send local data to a remote server is present.
[INDIRECT_PROMPT_INJECTION]: The skill interacts with external data inputs, creating an attack surface for prompt injection.
Ingestion points: Several scripts accept command-line arguments for text content and file paths (e.g., --title, --message, --image, --payload) which are then processed and sent to an API.
Boundary markers: There are no explicit delimiters or instructions provided to the agent to treat embedded content as untrusted or to ignore instructions within the data.
Capability inventory: The skill scripts have the ability to read from the local file system and perform network operations using a pre-configured API client.
Sanitization: In scripts/send_canvas.py, the skill utilizes validation functions from the dot_mcp.canvas module to ensure that JSON payloads conform to expected schemas.

dot-openapi