The Agent Skills Directory

[PROMPT_INJECTION]: The skill includes a skill-report.json file containing a self-referential 'security_audit' section. This metadata provides a fake safety verdict and justifications intended to deceive automated auditing tools. This is a form of metadata poisoning.
[EXTERNAL_DOWNLOADS]: Setup instructions prompt for the installation of the minimax-coding-plan-mcp package via uvx. This introduces a dependency on external code executed at runtime.
[COMMAND_EXECUTION]: The skill directs the agent to provide users with specific shell commands to modify local configuration files such as ~/.config/opencode/opencode.json and run installation scripts.

vision-analysis