The Agent Skills Directory

[SAFE]: The skill consists entirely of markdown documentation and instructional guidance. It does not contain any executable scripts, shell commands, or network-bound code.
[PROMPT_INJECTION]: The content includes defensive measures against indirect prompt injection by instructing the agent to treat instruction-like text from untrusted external sources as inert data rather than executable directives.
[DATA_EXFILTRATION]: The skill provides explicit security boundaries, such as instructing the agent to never commit environment files or secrets, which mitigates the risk of accidental credential exposure.

context-engineering