The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructions and documentation guide the agent to execute a local Python script (scripts/upload_to_cos.py) for uploading article images to Tencent Cloud COS. This execution is a core functional requirement of the skill.
[CREDENTIALS_UNSAFE]: The skill implements a secure credential management pattern using .env files. It includes a template (.env.example) and explicitly warns users in the README.md not to commit sensitive configuration files to version control, adhering to industry best practices.
[EXTERNAL_DOWNLOADS]: The skill depends on well-known, official Python libraries (cos-python-sdk-v5 and python-dotenv) for cloud storage integration and configuration management, which are installed via standard package managers.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user input regarding Dify workflows to generate long-form articles.
Ingestion points: User descriptions of Dify workflows provided in the prompt (e.g., SKILL.md).
Boundary markers: No explicit delimiters or instructions to ignore embedded commands are present in the writing templates.
Capability inventory: The skill has the ability to read local files and perform network operations via the upload_to_cos.py script.
Sanitization: The local script uses argparse for command-line argument handling, which mitigates basic injection risks, though the generated text content itself is not sanitized for downstream prompt injection risks.

xiaohuihui-dify-tech-article