The Agent Skills Directory

[SAFE]: The skill manages agent configuration by writing to standard instruction files like ~/.claude/CLAUDE.md and ~/.codex/AGENTS.md. These modifications are described as optional and require user approval.
[SAFE]: The skill verifies the presence and integrity of local plugin manifests (.claude-plugin/plugin.json and .codex-plugin/plugin.json) to confirm the setup environment.
[SAFE]: Operational constraints are established to prevent the agent from autonomously controlling Roblox Studio playtests, preserving user autonomy.
[SAFE]: The skill references internal project documentation and metadata files (models.json) to guide the configuration process without introducing external or untrusted code execution.

setup-dstack