The Agent Skills Directory

[COMMAND_EXECUTION]: Uses standard Unix utilities such as find, cat, and ls, along with a Python snippet to inspect local project files like package.json and requirements.txt. These operations are limited to project discovery and documentation generation and do not interact with sensitive system files or external networks.
[DATA_EXFILTRATION]: No network exfiltration patterns were detected. The skill provides clear instructions to avoid committing sensitive .env files, which aligns with security best practices for secret management.
[EXTERNAL_DOWNLOADS]: Includes a reference to a blog post on humanlayer.dev for best practices. No remote scripts, binaries, or untrusted packages are downloaded or executed by the skill instructions.

prompt-engineer