The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides example curl commands and Python scripts (requests.get) designed to interact with the Pollinations.ai API. These patterns demonstrate the capability to execute external network requests and process the resulting binary data.
[DATA_EXFILTRATION]: The skill facilitates sending user-provided prompts and generation parameters to the external domain image.pollinations.ai. While this is the primary purpose of the skill, it constitutes a transfer of local context to a third-party service.
[EXTERNAL_DOWNLOADS]: The skill provides patterns for downloading content from a remote server (https://image.pollinations.ai/prompt/) and saving it locally. The instructions include methods for both manual (curl) and automated (Python requests) retrieval of external binary assets.
[COMMAND_EXECUTION]: The Python code examples use with open(output_file, "wb") as f: f.write(response.content) and json.dump(metadata, f) to write files to the local filesystem. There is no evidence of path validation or sanitization, which creates a risk of overwriting arbitrary files if the output_file variable is influenced by untrusted input or malicious prompts.
[INDIRECT_PROMPT_INJECTION]: The skill processes external data (image responses) and generates local metadata files based on user prompts.
Ingestion points: API response content (binary image data) and user-supplied strings via Python's input or script arguments (inferred from variable usage).
Boundary markers: None present in the provided script templates.
Capability inventory: File writing (open().write()), Network requests (requests.get()).
Sanitization: No sanitization of file paths or URL parameters is shown in the examples.

pollinations-ai