The Agent Skills Directory

[DATA_EXFILTRATION]: The skill reads sensitive authentication tokens from local configuration files such as ~/.claude/.credentials.json, ~/.grok/auth.json, ~/.codex/auth.json, and ~/.config/cursor/auth.json.
On macOS, the skill executes the security utility to retrieve authentication secrets from the system Keychain.
These tokens are transmitted to the official usage endpoints of trusted and well-known providers, specifically api.anthropic.com and cli-chat-proxy.grok.com, to retrieve authorized usage statistics.
[COMMAND_EXECUTION]: The skill executes local binaries and system utilities to interact with provider tools and manage execution state.
It spawns CLI tools including codex, cursor-agent, claude, and grok from the system PATH or configured locations.
It utilizes the system script utility to emulate a TTY environment required by certain CLI tools to display usage info.
It executes the macOS security utility to facilitate Keychain credential retrieval.

t3-capacity