Anthropic adds "mods" to Claude Code: small TypeScript functions that change the agent's behavior
Prepared by Sintelligent with the help of SI. Publication time restored from the build log. . Updated: .
The agent is becoming a construction kit: its behavior can be replaced almost entirely, and built-in features are the same mods.
A mod can rewrite a prompt, block or repeat a tool call, approve or deny a permission request, hide secrets in output and change the interface; several mods stack into a chain. Mods are installed through plugins and the /plugin command and run without a sandbox with full access to the machine; Anthropic advises installing them only from trusted sources. On Team and Enterprise plans, the sec-default mod is planned to load by default, preventing permission denials from being bypassed.
Sources 1