The Agent Skills Directory

[COMMAND_EXECUTION]: The skill makes extensive use of subprocess.run to orchestrate its workflow. In generate.py, it calls other Python scripts (prompt_optimizer.py, generate_image.py, generate_video.py) using python3. Additionally, all script components (generate_image.py, generate_video.py, prompt_optimizer.py) execute the pass command-line utility to retrieve API keys from the local system's credential store.
[EXTERNAL_DOWNLOADS]: The skill performs network operations using the requests library to interact with an external API service at xingjiabiapi.com. It downloads media content (images and videos) from remote URLs returned by this API, including those hosted on s3.ffire.cc or other dynamically parsed links. These downloads are saved directly to the user's home directory under ~/clawd/output/.
[CREDENTIALS_UNSAFE]: The skill is designed to programmatically access the system's pass password manager to fetch the api/xingjiabiapi secret. While this avoids hardcoding keys, it establishes a pattern of the agent accessing sensitive system credentials during execution.
[DATA_EXFILTRATION]: User-provided prompts and descriptions are transmitted to the external domain xingjiabiapi.com for the purposes of prompt optimization and media generation. While this is the intended function of the skill, it involves sending potentially sensitive user data to a third-party service.

multimodal-gen