The Agent Skills Directory

[SAFE]: The skill provides standard operational instructions for an ImageKit MCP integration. It follows security best practices by explicitly instructing the agent never to request or process raw file bytes (Base64, data URIs) within the conversation context, which prevents both performance issues and potential data exposure risks.
[COMMAND_EXECUTION]: The skill describes a workflow involving a curlExample provided to the user. This is documented as a manual step for the user to perform in their own environment using a signed signature returned by a tool, rather than an instruction for the agent to execute commands autonomously or silently.
[INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied metadata such as filenames, tags, and custom fields. While this creates a minor exposure to indirect prompt injection via untrusted file names or metadata values, the instructions mandate use of specific structured tool calls and verification steps, which aligns with standard Digital Asset Management (DAM) operations.

upload-files