# TokenKick

For AI agents operating TokenKick, read `docs/AGENT_PLAYBOOK.md` first. Use
cached JSON reads for routine checks, avoid quota-consuming commands until the
user has approved the exact command, and do not invent scheduler behavior that
is not documented there.

Useful references:

- `docs/AGENT_PLAYBOOK.md`: safe headless agent operating guide.
- `docs/MCP.md`: first-party MCP server setup and safety contract.
- `docs/TOKENKICK_COMMANDS.md`: command reference.
- `docs/HOW_TOKENKICK_WORKS.md`: system behavior and state files.
- `docs/README.md`: user-facing overview.
