The Agent Skills Directory

[SAFE]: No malicious patterns, obfuscation, or data exfiltration attempts were detected across the skill's instructions, scripts, or reference files.
[COMMAND_EXECUTION]: The skill includes a local utility script, scripts/validate-metadata.py, which is invoked via python3 to validate skill metadata. The script's logic is safe, performing only string length and regex validation on the name and description fields without accessing the network or performing sensitive file operations.
[PROMPT_INJECTION]: The skill provides guidelines for agents on how to structure their own instructions, including how to avoid failure modes like "Negation" (telling the agent what not to do). These instructions are educational and intended to improve agent performance rather than bypass safety guardrails.

writing-skills