The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute a git log command to check for prior file reclassifications. This is a legitimate use of version control history to respect previous organizational decisions and does not involve any dangerous shell injection or privilege escalation.- [PROMPT_INJECTION]: An analysis of the instructions and reference materials revealed no attempts to bypass safety filters, override system instructions, or extract sensitive model information. The tone is pedagogical and aligned with documentation best practices.- [DATA_EXFILTRATION]: While the skill reads project documentation and configuration files, it does not contain any instructions or network operations that would transmit this data to external or untrusted domains. All suggested actions are local to the user's project environment.- [INDIRECT_PROMPT_INJECTION]: The skill has a surface for indirect prompt injection as it reads existing documentation files (Markdown, RST, AsciiDoc) which could contain malicious instructions. However, this is a necessary part of its documentation analysis function, and the skill does not interpolate this data into sensitive executable contexts in a way that would facilitate an attack.

diataxis