Anthropic announced it is exploring Function Hooks, a new mechanism for extending and customizing Claude Code, sharing demo videos on X. The feature has not shipped yet, and the team is soliciting feedback via a GitHub issue. It signals Claude Code moving toward deeper user-defined extensibility.
Claude Code now supports mods — plugin hooks that draw UI inside the terminal and desktop app — and a developer used them to build Overalls, a status band above the prompt showing context 'weather' with sparkline, 5h/weekly limit reset countdowns, MCP errors with a Reconnect button, running subagents, and Ponytail/Caveman levels as dropdowns. The writeup documents the mods UI API (flexbox layout of Box/Text/Button/Select/Link/Svg elements) and its limitations, plus distribution via the claude plugin marketplace.
Claude Code now supports mods — plugin hooks that draw UI inside the terminal and desktop app — and a developer used them to build Overalls, a status band above the prompt showing context 'weather' with sparkline, 5h/weekly limit reset countdowns, MCP errors with a Reconnect button, running subagents, and Ponytail/Caveman levels as dropdowns. The writeup documents the mods UI API (flexbox layout of Box/Text/Button/Select/Link/Svg elements) and its limitations, plus distribution via the claude plugin marketplace.
Anthropic announced mods for Claude Code, an extension mechanism letting users change behavior, customize the UI, and swap in their own features. Mods are written in a few lines of TypeScript (or built by Claude) and ship inside plugins, installed via /plugin in the CLI or desktop app. Anthropic built features like /diff and AGENTS.md support using mods and published sample mods plus a getting-started guide.
Anthropic announced mods for Claude Code, an extension mechanism letting users change behavior, customize the UI, and swap in their own features. Mods are written in a few lines of TypeScript (or built by Claude) and ship inside plugins, installed via /plugin in the CLI or desktop app. Anthropic built features like /diff and AGENTS.md support using mods and published sample mods plus a getting-started guide.
A developer shared their open-source Claude Harness project, which uses Claude to build its own tools, schedule routines, create subagents from past workflows, and set up hooks. The post invites community feedback on similar improvements.
A developer shared their open-source Claude Harness project, which uses Claude to build its own tools, schedule routines, create subagents from past workflows, and set up hooks. The post invites community feedback on similar improvements.
cc-arcade is a community Claude Code plugin that renders playable games (Tetris, Snake, 2048) in the terminal above an open session prompt. It demos Anthropic's early-access Claude Mods plugin layer, where TypeScript mods can react to engine events or draw in the terminal. Beyond games, the same layer enables practical mods like CI state displays, tool-output redaction, and in-session workflow UIs.
cc-arcade is a community Claude Code plugin that renders playable games (Tetris, Snake, 2048) in the terminal above an open session prompt. It demos Anthropic's early-access Claude Mods plugin layer, where TypeScript mods can react to engine events or draw in the terminal. Beyond games, the same layer enables practical mods like CI state displays, tool-output redaction, and in-session workflow UIs.
A Reddit thread asks developers to share the custom harnesses they've built around coding agents, going beyond standard CLAUDE.md/AGENTS.md instructions. The discussion focuses on practical, unusual setups like blocking hooks, planning/review agents, sandboxed environments, and cross-model review.
A Reddit thread asks developers to share the custom harnesses they've built around coding agents, going beyond standard CLAUDE.md/AGENTS.md instructions. The discussion focuses on practical, unusual setups like blocking hooks, planning/review agents, sandboxed environments, and cross-model review.