The Agent Skills Directory

[SAFE]: The skill provides benign instructional content focused on a research-driven development workflow. No evidence of prompt injection, data exfiltration, or obfuscation was found.
[EXTERNAL_DOWNLOADS]: The documentation references standard, well-known software packages from established registries (npm, PyPI) such as eslint, httpx, and zod. These are presented as examples of existing solutions to avoid redundant coding.
[COMMAND_EXECUTION]: The workflow suggests using common development tools like rg (ripgrep) to inspect the local repository for existing implementations, which is a legitimate and expected task in a coding environment.
[DATA_EXFILTRATION]: Instructions to check ~/.claude/settings.json and ~/.claude/skills/ are intended for the agent to determine its own available capabilities and configuration. There are no associated network operations to transmit this information externally.

search-first