A tiny pixel theater above your Claude Code prompt: little orange Claude acts out each tool call at a desk while a monitor shows the real command and Haiku narrates it in your request's language.

Claude Code mods, plugins, skills, hooks, status lines and MCP servers. Browse community builds with original sources and L0–L3 permission levels.
764 projectsA tiny pixel theater above your Claude Code prompt: little orange Claude acts out each tool call at a desk while a monitor shows the real command and Haiku narrates it in your request's language.

Advanced IPM compliance: confidentiality legend rule for every AI-made deliverable, checked before hand-off
work-settings Claude Code mod: context notifications, a branch band, a skills-list pane, and an agent-monitoring pane
Track what your AI coding tools cost
syrtis.nyanako.comVisit sponsor ↗A colorful little mascot in a pane that reacts to what you're doing, with short AI-written thoughts.
A Claude Code plugin that refuses or rewrites tool calls according to policy and, when observation is enabled, records hook events as database rows per lineage.
A Claude Code plugin that hooks Bash tool calls to block destructive commands, such as those that truncate output, detach processes, or mass-kill by name.
Progress bars with stage colors, ETA, subagent strips and automatic session titles
BazaarLink is an OpenAI-compatible AI API gateway: one key for 140+ AI models, pay as you go, with budget limits.
bazaarlink.aiVisit sponsor ↗A Claude Code plugin that wires revdiff into a floating Zellij pane and pipes diff annotations straight into the Claude Code session on each flush.
A Claude Code function-hook plugin implementing SKILL.state: an explicit, JSON-schema-validated execution state that gates non-exempt tool calls and replaces append-only conversation history with a compacted frame each turn, aimed at long-horizon skills.
A Claude Code mod that rewrites a prompt with Sonnet 5.5 at high effort and puts the result back in the prompt box, so you review it before sending. The rewrite follows the [Opus 5.5 prompting guide](https://claude.dev/blog/getting-the-most-out-of-opus-5-5/): the whole task in one message, a "done means" line for changes, when to stop and ask, and no "think step by step" or emphasis. It keeps the language of your prompt, and reads the working directory and the last few messages only to turn references like "that bug" into concrete names.
A Claude Code mod that blocks Claude from running scripts piped straight from the internet, including download-and-execute pipes, encoded PowerShell, and LOLBin downloaders, allowing safe patterns like curl -o and git clone.
Adds repo (and worktree, when in one), branch, PR, context and model to the end of the line under the prompt: ``` ▸▸ auto mode on (shift+tab to cycle) · new project / SHOP-142 · main (PR #142) · 100k/1000k (opus-5-5) ``` Branch and PR refresh at start, after each turn and every minute. PR needs `gh`.