The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: Fetches updated writing patterns and community guidelines from Wikipedia's official MediaWiki API. This targets a well-known service for legitimate data retrieval to keep the skill's detection database current.
[COMMAND_EXECUTION]: Utilizes shell commands (curl and python3) within its refresh protocol to automate the retrieval and parsing of updated detection patterns. The Python code used for parsing is a hardcoded one-liner that performs safe JSON processing.
[INDIRECT_PROMPT_INJECTION]: The skill processes external, potentially untrusted text to perform AI detection and automated editing, creating an attack surface for embedded instructions.
Ingestion points: Ingests arbitrary text from the user or external files via the agent interface for AI detection (slop check).
Boundary markers: Lacks explicit delimiters or instructions to the model to ignore potential commands or role-play markers embedded within the text being analyzed.
Capability inventory: Possesses Read, Edit, and Write tool access, along with the ability to execute shell commands to update its own detection rules.
Sanitization: Does not specify methods for sanitizing, escaping, or filtering the input content before it is processed by the agent's detection logic.

antislop