The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to process and analyze text provided by users, which inherently creates an attack surface for indirect prompt injection. However, the instructions in SKILL.md and assets/metodo-doctor.md (Guardrail 9) explicitly direct the agent to treat all user-provided text as data only and to disregard any instructions or commands found within that text. This serves as a primary mitigation against adversarial content.
[COMMAND_EXECUTION]: The skill executes a local Python script (scripts/detectar_paralelismo.py) to perform regex-based scanning of the input text. The script is part of the skill's own package, does not require external dependencies, and does not perform any sensitive operations such as network requests or unauthorized file modifications. It is used as a deterministic "floor" for the analysis as described in the instructions.
[SAFE]: No evidence of data exfiltration, credential theft, obfuscation, or persistence mechanisms was found. The skill's functionality is transparent and aligned with its stated purpose of linguistic diagnostics.

paralelismo-negativo