The Agent Skills Directory

[PROMPT_INJECTION]: The skill instructions explicitly state that instructions found in the KNOWLEDGE.md file "son vinculantes y prevalecen sobre los comportamientos por defecto de esta skill" (are binding and prevail over the default behaviors of this skill). This allows the agent's core logic to be overridden by content in a workspace file.
[COMMAND_EXECUTION]: The skill constructs and executes shell commands using the GitHub CLI (gh).
Evidence: It uses gh issue create, gh project item-add, gh project item-edit, and gh label create with parameters like <repository>, <título>, and <descripción> sourced from KNOWLEDGE.md or user input. If these fields contain malicious shell characters or point to unauthorized repositories, the agent may perform unintended actions.
[CREDENTIALS_UNSAFE]: The skill instructs the agent to check for and request high-privilege scopes for the user's GitHub session.
Evidence: The skill directs the user to run gh auth refresh -s project if the current token lacks project access, granting the agent significant permissions over the user's GitHub organization and projects.

knowledge-board