The Agent Skills Directory

[PROMPT_INJECTION]: The skill uses natural language instructions to adopt a Socratic persona. It does not include commands to override system instructions, bypass safety filters, or extract system prompts.
[DATA_EXFILTRATION]: No hardcoded credentials or unauthorized network operations were detected. The instruction to consult 'fuentes fiables' (reliable sources) to verify facts is a standard research function and does not target sensitive user data.
[COMMAND_EXECUTION]: The skill instructs the agent to 'explorar el código' (explore the code) when necessary to resolve questions. This is a functional capability intended for analyzing user-provided projects and does not involve arbitrary command execution or privilege escalation.
[PROMPT_INJECTION]: The skill has a surface for indirect prompt injection because it processes untrusted data (external code and web content). However, this is consistent with its primary purpose of analysis.
Ingestion points: Local code files and external web search results.
Boundary markers: Not explicitly defined in the instructions.
Capability inventory: The agent is permitted to read local files and perform web searches.
Sanitization: No specific filtering or sanitization steps are defined for the processed content.

charla-socratica