The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses shell commands to invoke Python scripts (image_create.py and publish.py) located in sibling directories within the same application suite to handle image generation and content publishing.\n- [CREDENTIALS_UNSAFE]: The skill is configured to read sensitive API keys and WeChat platform secrets (IMAGE_MODEL_API_KEY, WECHAT_1_APPID, WECHAT_1_APPSECRET) from a local file named aws.env. Storing credentials in a local environment file is a standard practice for secret management.\n- [INDIRECT_PROMPT_INJECTION]: The skill ingests user-provided themes and material images to generate prompts for external image models, creating a vulnerability surface where malicious data could influence agent behavior.\n
Ingestion points: User-provided themes, topics, and source images are ingested during the workflow's topic determination phase (Step 2).\n
Boundary markers: The documentation does not specify the use of delimiters or specific instructions to isolate untrusted user data within the generated prompts.\n
Capability inventory: The agent possesses capabilities for filesystem writes, shell command execution, and making network requests to third-party APIs.\n
Sanitization: There is no mention of input validation, filtering, or sanitization processes for the user-provided inputs before they are interpolated into prompts.\n- [EXTERNAL_DOWNLOADS]: The skill references its source code and dependency suite on the vendor's official GitHub repository and suggests installation via the clawhub tool.

aws-wechat-sticker