The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill requires the installation of the @runcomfy/cli package from the official npm registry. RunComfy is a well-known service providing hosted AI model infrastructure, and its official tools are considered safe for use.
[COMMAND_EXECUTION]: It invokes the runcomfy CLI to process image edits. The skill documentation demonstrates that inputs are passed as structured JSON, which is a secure method that prevents shell injection attacks.
[CREDENTIALS_UNSAFE]: Security best practices are documented for handling API tokens, including the use of environment variables or restricted configuration files (~/.config/runcomfy/token.json with 0600 permissions).
[DATA_EXFILTRATION]: Network activity is restricted to the official RunComfy API domains for model inference and result retrieval. No unauthorized data transmission or suspicious network patterns were detected.
[PROMPT_INJECTION]: The prompting guidelines provided are intended to improve the accuracy of image edits and do not contain any instructions aimed at bypassing model safety constraints or hijacking agent behavior.

flux-kontext