The Agent Skills Directory

[COMMAND_EXECUTION]: The skill contains instructions in SKILL.md for the agent to update itself by executing git clone and overwriting its local directory when prompted by the user. This operation targets the author's own repository and is a documented management feature.
[EXTERNAL_DOWNLOADS]: The skill identifies and references the author's GitHub repository (https://github.com/yzfly/CTO-Skills) for fetching updates and maintaining the skill's source files.
[PROMPT_INJECTION]: Static analysis identified potential prompt injection patterns in references/payload/methods/browser-testing-with-devtools.md. Upon review, these were determined to be examples within a security guide that instructs the agent on how to recognize and resist injection attempts in untrusted browser data. This is a beneficial safety instruction and a false positive for malicious intent.
[SAFE]: The skill incorporates robust defensive coding and security guidelines for the agent to follow when interacting with external data and browsers, explicitly warning against interpreting data as instructions and accessing sensitive credentials.