Excalidraw

Excalidraw files are deceptively expensive,a single diagram can burn 4k-22k tokens because the JSON is packed with visual metadata (positioning, styling, roughness values) while actual content is maybe 10% of the file. This skill enforces a hard rule: main agents never read .excalidraw files directly, always delegate to subagents instead. The pattern isolates context pollution so you can work with multiple diagrams without torching your budget. Real example from the source: reading 7 project diagrams directly costs 67k tokens (33% of your context), delegating costs about 2k. Worth using even for "small" files or "quick checks" since the smallest diagrams still hit 4k tokens and the temptation to peek is strong.

skills.sh installs: 804
GitHub stars: 33.2K
last updated: Aug 21, 2026

View on GitHub

Install

$ npx -y skills add davila7/claude-code-templates --skill excalidraw --agent claude-code

Installs into .claude/skills of the current project.

Files

Excalidraw Subagent Delegation

Overview

Core principle: Main agents NEVER read Excalidraw files directly. Always delegate to subagents to isolate context consumption.

Excalidraw files are JSON with high token cost but low information density. Single files range from 4k-22k tokens (largest can exceed read tool limits). Reading multiple diagrams quickly exhausts context budget (7 files = 67k tokens = 33% of budget).

The Problem

Excalidraw JSON structure:

Each shape has 20+ properties (x, y, width, height, strokeColor, seed, version, etc.)
Most properties are visual metadata (positioning, styling, roughness)
Actual content: text labels and element relationships (<10% of file)
Signal-to-noise ratio is extremely low

Example: 14-element diagram = 596 lines, 16K, ~4k tokens. 79-element diagram = 2,916 lines, 88K, ~22k tokens (exceeds read limit).

When to Use

SKILL.md on GitHub

Details

Repository: github.com/davila7/claude-code-templates
Categories: Code Review & Quality, Data Science & ML
First seen: May 16, 2026

evalulu-eva/eva-skillEva-skill 2.2.8 的极薄路由入口。仅在用户调用 /eva、点名 Eva 子入口，提出 Eva 的思考、人群识别、短视频或非虚构文章、发布前审核、学习、Brief、复盘、Lens、Memory、Link、新手教程，或在 Eva 上下文中询问“下一步怎么走、先用哪个功能、入口排序、给我一个工作流”时使用；也处799skills.sh installs

eachlabs-image-editeachlabs/skillsEdit, transform, upscale, and enhance images using EachLabs AI models. Supports image editing, style transfer, background removal, upscaling, inpainting, face s792skills.sh installs

evaluation-rubricslyndonkl/claudeDesigns structured scoring tools with explicit criteria, performance scales, and descriptors for consistent, transparent quality assessment. Use when need quali790skills.sh installs

clangmohitmishra786/low-level-dev-skillsClang/LLVM compiler skill for C/C++ projects. Use when working with clang or clang++ for diagnostics, sanitizer instrumentation, optimization remarks, static an789skills.sh installs

dockerfile-validatorakin-ozer/cc-devops-skillsValidate, lint, audit, or scan a Dockerfile for security and best practices.786skills.sh installs

scaffold-nextjsmblode/agent-skillsScaffolds a production-ready Next.js turborepo end to end. Runs create-next-app with TypeScript 7, Tailwind CSS, React Compiler, and Cache Components, sets up s785skills.sh installs