The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted curriculum data provided via the kud_json and lts_json inputs in SKILL.md. While boundary markers for this data are limited to standard template variable placeholders, the skill has no privileged capabilities such as network access, file writing, or command execution. Furthermore, the 'Source-voice preservation rule' acts as a sanitization mechanism by strictly forbidding the agent from modifying the input content or interpreting it as instructions.
[NO_CODE]: The skill consists entirely of markdown documentation and YAML metadata. It does not include any executable scripts, binaries, or external code dependencies.
[SAFE]: No patterns of malicious behavior, such as credential theft, obfuscation, or unauthorized access, were identified. The skill's operations are limited to data transformation and metadata enrichment within its defined instructional scope.

developmental-band-translator