The Agent Skills Directory

[PROMPT_INJECTION]: The skill exhibits a surface for indirect prompt injection because it is designed to ingest and process data from potentially untrusted external sources.
Ingestion points: The skill accepts user-provided text (materialText), file uploads (materialFile), and external URLs (materialUrl) as inputs, as defined in references/form-schema.json.
Boundary markers: No explicit boundary markers or instructions to ignore embedded directions are present in the provided instructions or scripts.
Capability inventory: The skill's primary capability is performing authenticated network requests to the ai-skills.ai API via scripts/run.py.
Sanitization: There is no evidence of input sanitization or validation within the skill's client-side components.
[DATA_EXFILTRATION]: The skill transmits user-provided content to the vendor's infrastructure at ai-skills.ai. This behavior is the intended primary function of the skill and targets the official domain of the author. Authentication is handled securely using an API key (AISKILLS_API_KEY) retrieved from environment variables.
[COMMAND_EXECUTION]: The execution script scripts/run.py uses the standard Python urllib library for API communication. It does not perform any unsafe shell command execution or subprocess spawning.

ai-social