Official launch: you can now mod Claude Code
Change how it behaves, customize the UI, swap in your own features — a few lines of TypeScript, or let Claude build it.
Claude Code mods, plugins, skills, hooks, status lines and MCP servers. Browse community builds with original sources and L0–L3 permission levels.
Launched: · Last updated:
Change how it behaves, customize the UI, swap in your own features — a few lines of TypeScript, or let Claude build it.
Additionally added mods for claude code. It's not just a dashboard—it's visible in real time inside claude code too. When lens is enabled, a line appears just below the input field, showing the cache hit rate of the latest request, the hit trend of recent requests, the time remaining until cache expiry (⏱ 3:12), and current context usage, all updating as each request completes. If the cache has less than 1 minute left, it turns yellow so you can decide whether to continue the conversation before it expires. In long sessions, this one timing decision can make the next request cost dozens of times more or less.
A Claude Code mod "Kirari" made in 90 minutes ✧ It appears only while Claude is thinking, showing Petit-tya (M5Stack) like a window. It makes sounds when touched. Plays with Kani-chan. Not flashy, but accompanies everyday life. #ClaudeBuildNight #claudeMods #Petit_Ones #ClaudeModsHack https://t.co/i3fsx66qeJ
18 AI agents. 7 repos. Now none of them can push to main, touch a secret, or wander into another repo. Claude Code Mods made that possible. 🧵
If you're working on something where you want to keep the fine details, before you leave your seat, either instruct what to keep with /compact focus on ... or add "idleCompaction": false to settings.json (this works in v2.1.290 and later). Long sessions, leaving your seat with them still open—doesn't that happen a lot? https://code.claude.com/docs/en/settings-reference#idlecompaction
Track what your AI coding tools cost
syrtis.nyanako.comVisit sponsor ↗Tired of AI-generated UIs that all look the same? Here are 10 GitHub repos that turn Claude Code and Codex into elite UI designers: 1. Superdesign Skill https://github.com/superdesigndev/superdesign-skill 2. Better Design https://github.com/marvkr/better-design
Collects the work progress of multiple sessions by work unit and shows it as a single line above the input box and in the /progress panel. Draws steps as connected nodes, and keeps track of remaining work after completion and unclosed sub-agent records.
Progress Ring — see how far along Claude Code is in a long plan: a ring band above the prompt with percentage and current step, a /progress panel across chats, and milestone rows.

In a Claude Code Mod, I displayed session task progress💡 For now, is this kind of information enough?🤔 ・Purpose and goals ・Current status (whether the AI is handling it or it's your turn) ・Next task ・Usage and remaining context window Including the design (this color lol) I don't like it, so I'll keep modifying it😂 https://t.co/pxUVIU4J6N
claude code 2.1.300 is worth your update, and the reason is mods. the headline change: agent.spawn hooks can now rewrite a workflow agent's prompt and model at spawn time. before, a hook could only refuse the agent. that turns routing into code: cheap model for doc sweeps, strong model for the actual fix. also in: prompt.change fires on non-typed prompt changes (history recall, undo, plugin fill), /advisor lands on claude platform on aws, and otel spans now carry genai semantic-convention attr
One key for 140+ AI models
bazaarlink.aiVisit sponsor ↗A CLAUDE CODE MOD CAN READ YOUR API KEYS AND APPROVE COMMANDS BEFORE YOU SEE THEM. HERE'S HOW TO CHECK ONE IN 30 SECONDS.
On October 14 and 15, Claude Code v2.1.300 and v2.1.301 were released in quick succession. In v2.1.300 (88 changes), Mods...
Based on a measured report on Zenn, this post explains that about 98% of Claude Code's cost is caused by re-reading conversation history, and introduces three money-saving techniques: resetting sessions, splitting roles, and consolidating tool calls.