The Agent Skills Directory

[SAFE]: The skill provides documentation for the weknora CLI tool, using appropriate placeholders and standard environment variables for configuration.\n- [COMMAND_EXECUTION]: The instructions describe shell commands for tool interaction. These are limited to the specific CLI tool (weknora) and follow common patterns for agent-based tool use.\n- [PROMPT_INJECTION]: The skill contains explicit safety instructions that prevent the agent from performing destructive actions without human confirmation, specifically prohibiting the automatic use of the -y flag.\n- [SAFE]: Authentication instructions favor stateless environment variables and stdin piping for tokens, which are security best practices for handling credentials in automated environments.\n- [PROMPT_INJECTION]: The skill defines patterns for processing server-side data from RAG search and chat tools. While this creates a potential surface for indirect prompt injection, it is managed through structured exit codes and JSON output handling instructions.

weknora-shared