The Agent Skills Directory

[PROMPT_INJECTION]: The skill's instructions focus on logical reasoning and fact-checking. No patterns attempting to bypass safety filters or override system instructions were detected.
[DATA_EXFILTRATION]: No evidence of unauthorized data access or exfiltration patterns, such as sensitive file paths or hardcoded credentials, was found.
[COMMAND_EXECUTION]: The provided files do not contain shell commands, binary execution, or any logic for invoking subprocesses.
[EXTERNAL_DOWNLOADS]: The skill mentions external references to technical standards (IETF) and documentation from major technology providers (Anthropic, OpenAI, Google) for educational purposes. These links point to well-known domains and do not involve executable payloads.
[NO_CODE]: This is a documentation-based skill consisting entirely of Markdown and JSON files; it includes no Python, JavaScript, or shell scripts.
[SAFE]: A comprehensive analysis across all threat categories confirms that the skill is purely instructional and poses no security risk to the agent environment.

epistemic-grounding