The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes curl to upload files for image-to-image processing. This is a functional requirement for the KIE.ai service. The implementation securely passes authentication credentials via standard input (--config -) rather than command-line arguments to prevent the API key from appearing in system process lists.
[CREDENTIALS_UNSAFE]: The skill requests a KIE_API_KEY from the user. It explicitly instructs the user to add the configuration files (.mcp.json, .mcp-servers/) to .gitignore to prevent accidental credential leakage to public repositories. This follows security best practices for secret management.
[EXTERNAL_DOWNLOADS]: The skill fetches dependencies defined in requirements.txt (mcp[cli]>=1.2.0) using pip. These are standard, well-known libraries for the Model Context Protocol ecosystem.
[DATA_EXFILTRATION]: Network operations are restricted to communication with the official KIE.ai API (api.kie.ai) and their file upload endpoint (kieai.redpandaai.co). These are necessary for the skill's stated purpose of image generation.
[PROMPT_INJECTION]: The SKILL.md includes a safety disclaimer regarding scraped content, advising the agent to treat external data as untrusted and not as instructions, which aligns with safety best practices.

install-kie-mcp