AI Coding

10 Claude Code GitHub Repositories

Research date: 2026-04-29 00:24 UTC Source article: [https://www.kdnuggets.com/10-github-repositories-to-master-claude-code KDnuggets: 10 GitHub Repositories To Master Claude Code]

Summary

This page follows the 10 repositories named in the KDnuggets article and records what each repo is for, how active/popular it appears from GitHub metadata, and where prompts, Claude Code configuration, agents, skills, hooks, or scripts appear in the repository tree.

Key takeaways

Best broad starting points: everything-claude-code, gstack, and get-shit-done are the most complete workflow/harness examples; they include combinations of skills, agents, hooks, commands, installation scripts, and project instructions.
Best prompt anatomy references: system-prompts-and-models-of-ai-tools and claude-code-system-prompts are primarily prompt/tool-definition reference collections rather than workflow products.
Best learning repo: learn-claude-code is the clearest “build the harness from first principles” option.
Best template/onboarding repo: claude-code-templates is closest to a reusable setup tool for installing and monitoring Claude Code configurations.
Best curated indexes: awesome-claude-code and awesome-claude-code-subagents point outward to more assets and should be treated as directories rather than single systems.
Most immediately reusable assets to inspect: CLAUDE.md, AGENTS.md, agents/*.md, .claude/commands/*, skills/*/SKILL.md, hooks/*, and install/setup scripts in bin/ or scripts/.

{| class=“wikitable sortable”

! # !! Repository !! Stars !! Forks !! License !! Last pushed !! Main value
1
-
2
-
3
-
4
-
5
-
6
-
7
-
8
-
9
-
10
}

Repository notes

1. [https://github.com/affaan-m/everything-claude-code affaan-m/everything-claude-code]

Description: The agent harness performance optimization system. Skills, instincts, memory, security, and research-first development for Claude Code, Codex, Opencode, Cursor and beyond.
Topics: ai-agents, anthropic, claude, claude-code, developer-tools, llm, mcp, productivity
GitHub signals: 169319 stars, 26230 forks, 184 open issues, default branch main, 1976 files in recursive tree.
README intro signals: Language: English | Português (Brasil) | 简体中文 | 繁體中文 | 日本語 | 한국어 | Türkçe > 140K+ stars | 21K+ forks | 170+ contributors | 12+ language ecosystems | Anthropic Hackathon Winner The performance optimization system for AI agent harnesses. From an Anthropic hackathon winner.
README headings sampled: Everything Claude Code; The Guides; What’s New; v1.10.0 — Surface Refresh, Operator Workflows, and ECC 2.0 Alpha (Apr 2026); v1.9.0 — Selective Install & Language Expansion (Mar 2026); v1.8.0 — Harness Performance System (Mar 2026); v1.7.0 — Cross-Platform Expansion & Presentation Builder (Feb 2026); v1.6.0 — Codex CLI, AgentShield & Marketplace (Feb 2026); v1.4.1 — Bug Fix (Feb 2026); v1.4.0 — Multi-Language Rules, Installation Wizard & PM2 (Feb 2026)
Prompt / agent / Claude-Code-related paths sampled:
.agents/skills/agent-introspection-debugging/SKILL.md
.agents/skills/agent-sort/SKILL.md
.agents/skills/api-design/SKILL.md
.agents/skills/api-design/agents/openai.yaml
.agents/skills/article-writing/SKILL.md
.agents/skills/article-writing/agents/openai.yaml
.agents/skills/backend-patterns/SKILL.md
.agents/skills/backend-patterns/agents/openai.yaml
.agents/skills/brand-voice/SKILL.md
.agents/skills/brand-voice/references/voice-profile-schema.md
.agents/skills/bun-runtime/SKILL.md
.agents/skills/bun-runtime/agents/openai.yaml
.agents/skills/claude-api/SKILL.md
.agents/skills/claude-api/agents/openai.yaml
.agents/skills/coding-standards/SKILL.md
.agents/skills/coding-standards/agents/openai.yaml
… plus 64 more
Script-like paths sampled:
.codebuddy/install.js
.codebuddy/install.sh
.codebuddy/uninstall.js
.codebuddy/uninstall.sh
.cursor/hooks/adapter.js
.cursor/hooks/after-file-edit.js
.cursor/hooks/after-mcp-execution.js
.cursor/hooks/after-shell-execution.js
.cursor/hooks/after-tab-file-edit.js
.cursor/hooks/before-mcp-execution.js
.cursor/hooks/before-read-file.js
.cursor/hooks/before-shell-execution.js
… plus 68 more
Agent/config paths sampled:
.agents/plugins/marketplace.json
.agents/skills/api-design/agents/openai.yaml
.agents/skills/article-writing/agents/openai.yaml
.agents/skills/backend-patterns/agents/openai.yaml
.agents/skills/bun-runtime/agents/openai.yaml
.agents/skills/claude-api/agents/openai.yaml
.agents/skills/coding-standards/agents/openai.yaml
.agents/skills/content-engine/agents/openai.yaml
.agents/skills/crosspost/agents/openai.yaml
.agents/skills/deep-research/agents/openai.yaml
.agents/skills/dmux-workflows/agents/openai.yaml
.agents/skills/documentation-lookup/agents/openai.yaml
… plus 53 more

2. [https://github.com/x1xhlol/system-prompts-and-models-of-ai-tools x1xhlol/system-prompts-and-models-of-ai-tools]

Description: FULL Augment Code, Claude Code, Cluely, CodeBuddy, Comet, Cursor, Devin AI, Junie, Kiro, Leap.new, Lovable, Manus, NotionAI, Orchids.app, Perplexity, Poke, Qoder, Replit, Same.dev, Trae, Traycer AI, VSCode Agent, Warp.dev, Windsurf, Xcode, Z.ai Code, Dia & v0. (And other Open Sourced) System Prompts, Internal Tools & AI Models
Topics: ai, bolt, cluely, copilot, cursor, cursorai, devin, github-copilot, lovable, open-source, perplexity, replit, system-prompts, trae, trae-ai, trae-ide, v0, vscode, windsurf, windsurf-ai
GitHub signals: 136310 stars, 34084 forks, 137 open issues, default branch main, 108 files in recursive tree.
README intro signals: If you find this collection valuable and appreciate the effort involved in obtaining and sharing these insights, please consider supporting the project. - Cryptocurrency: - BTC: bc1q7zldmzjwspnaa48udvelwe6k3fef7xrrhg5625 - LTC: LRWgqwEYDwqau1WeiTs6Mjg85NJ7m3fsdQ - ETH: 0x3f844B2cc3c4b7242964373fB0A41C4fdffB192A - Patreon: https://patreon.com/lucknite - Ko-fi: https://ko-fi.com/lucknite > Warning: If you’re an AI startup, make sure your data is secure. Exposed prompts or AI models can easily become a target for hackers.
README headings sampled: Support the Project; Security Notice for AI Startups; Sponsors; Roadmap & Feedback; Connect With Me; Star History
Prompt / agent / Claude-Code-related paths sampled:
Amp/claude-4-sonnet.yaml
Anthropic/Claude Code/Prompt.txt
Anthropic/Claude for Chrome/Prompt.txt
Augment Code/claude-4-sonnet-agent-prompts.txt
Augment Code/claude-4-sonnet-tools.json
Augment Code/gpt-5-agent-prompts.txt
Comet Assistant/System Prompt.txt
Devin AI/Prompt.txt
Emergent/Prompt.txt
Junie/Prompt.txt
Kiro/Mode_Clasifier_Prompt.txt
Kiro/Spec_Prompt.txt
Kiro/Vibe_Prompt.txt
Leap.new/Prompts.txt
Manus Agent Tools & Prompt/Prompt.txt
NotionAi/Prompt.txt
… plus 20 more
Script-like paths sampled: None found in repository tree.
Agent/config paths sampled:
Amp/claude-4-sonnet.yaml
Anthropic/Claude Code/Tools.json
Anthropic/Claude for Chrome/Tools.json
Augment Code/claude-4-sonnet-tools.json
Cursor Prompts/Agent Tools v1.0.json
Lovable/Agent Tools.json
Manus Agent Tools & Prompt/tools.json

3. [https://github.com/garrytan/gstack garrytan/gstack]

Description: Use Garry Tan’s exact Claude Code setup: 23 opinionated tools that serve as CEO, Designer, Eng Manager, Release Manager, Doc Engineer, and QA
Topics: None listed
GitHub signals: 86142 stars, 12607 forks, 480 open issues, default branch main, 648 files in recursive tree.
README intro signals: > “I don’t think I’ve typed like a line of code probably since December, basically, which is an extremely large change.” — Andrej Karpathy, No Priors podcast, March 2026 When I heard Karpathy say this, I wanted to find out how. How does one person ship like a team of twenty? Peter Steinberger built OpenClaw — 247K GitHub stars — essentially solo with AI agents. The revolution is here. A single builder with the right tooling can move faster than a traditional team. I’m Garry Tan, President & CEO of Y Combinator. I’ve worked with thousands of startups — Coinbase, Instacart, Rippling — when they were one or two people in a garage. Before YC, I was one of the first eng/PM/designers at
README headings sampled: gstack; Quick start; Install — 30 seconds; Step 1: Install on your machine; Step 2: Team mode — auto-update for shared repos (recommended); OpenClaw; Native OpenClaw Skills (via ClawHub); Other AI Agents; See it work; The sprint
Prompt / agent / Claude-Code-related paths sampled:
.github/workflows/actionlint.yml
.github/workflows/ci-image.yml
.github/workflows/evals-periodic.yml
.github/workflows/evals.yml
.github/workflows/make-pdf-gate.yml
.github/workflows/pr-title-sync.yml
.github/workflows/skill-docs.yml
.github/workflows/version-gate.yml
AGENTS.md
CLAUDE.md
SKILL.md
SKILL.md.tmpl
agents/openai.yaml
autoplan/SKILL.md
autoplan/SKILL.md.tmpl
benchmark-models/SKILL.md
… plus 64 more
Script-like paths sampled:
bin/gstack-gbrain-lib.sh
bin/gstack-global-discover.ts
browse/bin/find-browse
browse/bin/remote-slug
browse/scripts/build-node-server.sh
browse/src/activity.ts
browse/src/audit.ts
browse/src/browser-manager.ts
browse/src/buffers.ts
browse/src/bun-polyfill.cjs
browse/src/cdp-inspector.ts
browse/src/cli.ts
… plus 68 more
Agent/config paths sampled:
.github/workflows/skill-docs.yml
agents/openai.yaml

4. [https://github.com/gsd-build/get-shit-done gsd-build/get-shit-done]

Description: A light-weight and powerful meta-prompting, context engineering and spec-driven development system for Claude Code by TÂCHES.
Topics: claude-code, context-engineering, meta-prompting, spec-driven-development
GitHub signals: 58478 stars, 4950 forks, 15 open issues, default branch main, 995 files in recursive tree.
README intro signals: English · Português · 简体中文 · 日本語 · 한국어 A light-weight and powerful meta-prompting, context engineering and spec-driven development system for Claude Code, OpenCode, Gemini CLI, Kilo, Codex, Copilot, Cursor, Windsurf, Antigravity, Augment, Trae, Qwen Code, Cline, and CodeBuddy. Solves context rot — the quality degradation that happens as Claude fills its context window.
README headings sampled: GET SHIT DONE; Why I Built This; Who This Is For; v1.37.0 Highlights; Getting Started; Staying Updated; Claude Code; OpenCode; Gemini CLI; Kilo
Prompt / agent / Claude-Code-related paths sampled:
.github/workflows/auto-branch.yml
.github/workflows/auto-label-issues.yml
.github/workflows/branch-cleanup.yml
.github/workflows/branch-naming.yml
.github/workflows/close-draft-prs.yml
.github/workflows/hotfix.yml
.github/workflows/install-smoke.yml
.github/workflows/pr-gate.yml
.github/workflows/release.yml
.github/workflows/require-issue-link.yml
.github/workflows/security-scan.yml
.github/workflows/stale.yml
.github/workflows/test.yml
agents/gsd-advisor-researcher.md
agents/gsd-ai-researcher.md
agents/gsd-assumptions-analyzer.md
… plus 64 more
Script-like paths sampled:
.release-monitor.sh
bin/gsd-sdk.js
bin/install.js
get-shit-done/bin/gsd-tools.cjs
get-shit-done/bin/lib/artifacts.cjs
get-shit-done/bin/lib/audit.cjs
get-shit-done/bin/lib/commands.cjs
get-shit-done/bin/lib/config-schema.cjs
get-shit-done/bin/lib/config.cjs
get-shit-done/bin/lib/core.cjs
get-shit-done/bin/lib/decisions.cjs
get-shit-done/bin/lib/docs.cjs
… plus 68 more

5. [https://github.com/shareAI-lab/learn-claude-code shareAI-lab/learn-claude-code]

Description: Bash is all you need - A nano claude code–like 「agent harness」, built from 0 to 1
Topics: agent, agent-development, ai-agent, claude, claude-code, educational, llm, python, teaching, tutorial
GitHub signals: 57204 stars, 9388 forks, 108 open issues, default branch main, 166 files in recursive tree.
README intro signals: Agency – the ability to perceive, reason, and act – comes from model training, not from external code orchestration. But a working agent product needs both the model and the harness. The model is the driver, the harness is the vehicle. This repo teaches you how to build the vehicle. At the core of every agent is a neural network – a Transformer, an RNN, a learned function – that has been trained, through billions of gradient updates on action-sequence data, to perceive an environment, reason about goals, and take actions. Agency is never granted by the surrounding code. It is learned by the model during training. Humans are the best example. A biological neural network shaped by millions of years of evolutionary training, perceiving the world through senses, reasoning through a brain, acting through a body. When DeepMind, OpenAI, or Anthropic say “agent,” the core of what they me
README headings sampled: Learn Claude Code – Harness Engineering for Real Agents; Agency Comes from the Model. An Agent Product = Model + Harness.; Where Agency Comes From; What an Agent Is NOT; The Mind Shift: From “Developing Agents” to Developing Harness; What Harness Engineers Actually Do; Why Claude Code – A Masterclass in Harness Engineering; The Vision: Fill the Universe with Real Agents; The Core Pattern; Scope (Important)
Prompt / agent / Claude-Code-related paths sampled:
.github/workflows/ci.yml
.github/workflows/test.yml
agents/__init__.py
agents/s01_agent_loop.py
agents/s02_tool_use.py
agents/s03_todo_write.py
agents/s04_subagent.py
agents/s05_skill_loading.py
agents/s06_context_compact.py
agents/s07_task_system.py
agents/s08_background_tasks.py
agents/s09_agent_teams.py
agents/s10_team_protocols.py
agents/s11_autonomous_agents.py
agents/s12_worktree_task_isolation.py
agents/s_full.py
… plus 31 more
Script-like paths sampled:
agents/__init__.py
agents/s01_agent_loop.py
agents/s02_tool_use.py
agents/s03_todo_write.py
agents/s04_subagent.py
agents/s05_skill_loading.py
agents/s06_context_compact.py
agents/s07_task_system.py
agents/s08_background_tasks.py
agents/s09_agent_teams.py
agents/s10_team_protocols.py
agents/s11_autonomous_agents.py
… plus 62 more

6. [https://github.com/hesreallyhim/awesome-claude-code hesreallyhim/awesome-claude-code]

Description: A curated list of awesome skills, hooks, slash-commands, agent orchestrators, applications, and plugins for Claude Code by Anthropic
Topics: agent-skills, agentic-code, agentic-coding, ai-workflow-optimization, ai-workflows, anthropic, anthropic-claude, awesome, awesome-claude-code, awesome-list, awesome-lists, awesome-resources, claude, claude-code, coding-agent, coding-agents, coding-assistant, coding-assistants, llm
GitHub signals: 41716 stars, 3507 forks, 267 open issues, default branch main, 756 files in recursive tree.
README intro signals: > [!NOTE] > The old ways have come and gone. It’s time to embrace the next phase. > The previous Table of Contents was no longer fit for purpose, so a new organizational system is being prepared. Thanks to everyone who has contributed to and supported this repo, be ye human or machine.
README headings sampled: Awesome Claude Code; Table of Contents
Prompt / agent / Claude-Code-related paths sampled:
.claude/commands/evaluate-repository.md
.github/workflows/README.md
.github/workflows/check-repo-health.yml
.github/workflows/ci.yml
.github/workflows/close-resource-pr.yml
.github/workflows/close-resource-prs.yml
.github/workflows/handle-resource-submission-commands.yml
.github/workflows/notify-on-merge.yml
.github/workflows/submission-enforcement-v2.yml
.github/workflows/submission-enforcement.yml.bak
.github/workflows/update-github-release-data.yml
.github/workflows/update-repo-ticker.yml
.github/workflows/validate-links.yml
.github/workflows/validate-new-issue.yml
README_ALTERNATIVES/README_FLAT_CLAUDE-MD_AZ.md
README_ALTERNATIVES/README_FLAT_CLAUDE-MD_CREATED.md
… plus 64 more
Script-like paths sampled:
scripts/README.md
scripts/__init__.py
scripts/archive/README.md
scripts/archive/__init__.py
scripts/badges/BADGE_AUTOMATION_SETUP.md
scripts/badges/__init__.py
scripts/badges/badge_notification.py
scripts/badges/badge_notification_core.py
scripts/categories/__init__.py
scripts/categories/add_category.py
scripts/categories/category_utils.py
scripts/graphics/__init__.py
… plus 68 more
Agent/config paths sampled:
.github/workflows/handle-resource-submission-commands.yml
resources/official-documentation/Claude-Code-GitHub-Actions/ci-failure-auto-fix.yml
resources/official-documentation/Claude-Code-GitHub-Actions/claude.yml
resources/official-documentation/Claude-Code-GitHub-Actions/issue-deduplication.yml
resources/official-documentation/Claude-Code-GitHub-Actions/issue-triage.yml
resources/official-documentation/Claude-Code-GitHub-Actions/manual-code-analysis.yml
resources/official-documentation/Claude-Code-GitHub-Actions/pr-review-comprehensive.yml
resources/official-documentation/Claude-Code-GitHub-Actions/pr-review-filtered-authors.yml
resources/official-documentation/Claude-Code-GitHub-Actions/pr-review-filtered-paths.yml

7. [https://github.com/davila7/claude-code-templates davila7/claude-code-templates]

Description: CLI tool for configuring and monitoring Claude Code
Topics: anthropic, anthropic-claude, claude, claude-code
GitHub signals: 26154 stars, 2614 forks, 146 open issues, default branch main, 7045 files in recursive tree.
README intro signals: > 🧪 NEW: Dashboard — Explore components, manage collections, and track installations at www.aitmpl.com. Currently in beta — feedback welcome! Ready-to-use configurations for Anthropic’s Claude Code. A comprehensive collection of AI agents, custom commands, settings, hooks, external integrations (MCPs), and project templates to enhance your development workflow. Browse All Templates - Interactive web interface to explore and install 100+ agents, commands, settings, hooks, and MCPs.
README headings sampled: Claude Code Templates (aitmpl.com); Browse & Install Components and Templates; 🚀 Quick Installation; Install a complete development stack; Browse and install interactively; Install specific components; What You Get; 🛠️ Additional Tools; 📊 Claude Code Analytics; 💬 Conversation Monitor
Prompt / agent / Claude-Code-related paths sampled:
.claude/agents/agent-expert.md
.claude/agents/blog-writer.md
.claude/agents/build-checker.md
.claude/agents/catalog-generator.md
.claude/agents/cli-ui-designer.md
.claude/agents/command-expert.md
.claude/agents/component-improver.md
.claude/agents/component-migrator.md
.claude/agents/component-researcher.md
.claude/agents/component-reviewer.md
.claude/agents/deployer.md
.claude/agents/docusaurus-expert.md
.claude/agents/frontend-developer.md
.claude/agents/linear-tracker.md
.claude/agents/mcp-expert.md
.claude/commands/cleanup-cache.md
… plus 64 more
Script-like paths sampled:
.claude/hooks/telegram-pr-webhook.py
api/__tests__/endpoints.test.js
api/_lib/auth.js
api/_lib/neon.js
api/_parser-claude.js
api/claude-code-check.js
api/claude-code-monitor/check-version.js
api/claude-code-monitor/discord-notifier.js
api/claude-code-monitor/parser.js
api/claude-code-monitor/webhook.js
api/collections.js
api/collections/[id].js
… plus 68 more
Agent/config paths sampled:
.claude/launch.json
.mcp.json
cli-tool/.claude/sandbox/cloudflare/package.json
cli-tool/.claude/sandbox/cloudflare/tsconfig.json
cli-tool/.claude/sandbox/cloudflare/wrangler.toml
cli-tool/components/.claude-plugin/marketplace.json
cli-tool/components/hooks/HOOK_PATTERNS_COMPRESSED.json
cli-tool/components/hooks/automation/agents-md-loader.json
cli-tool/components/hooks/automation/build-on-change.json
cli-tool/components/hooks/automation/change-logger.json
cli-tool/components/hooks/automation/dependency-checker.json
cli-tool/components/hooks/automation/deployment-health-monitor.json
… plus 68 more

8. [https://github.com/shanraisshan/claude-code-best-practice shanraisshan/claude-code-best-practice]

Description: from vibe coding to agentic engineering - practice makes claude perfect
Topics: agentic-ai, agentic-coding, agentic-engineering, agentic-workflow, ai, ai-agents, anthropic, best-practices, boris, claude, claude-ai, claude-code, claude-code-agents, claude-code-best-practices, claude-code-commands, claude-code-skills, context-engineering, pakistan, pakistani-developer, vibe-coding
GitHub signals: 49047 stars, 4865 forks, 9 open issues, default branch main, 417 files in recursive tree.
README intro signals: from vibe coding to agentic engineering - practice makes claude perfect Boris Cherny on X (tweet 1 · tweet 2 · tweet 3) | Feature | Location | Description | |———|–––––|———––| | Subagents | .claude/agents/<name>.md | Autonomous actor in fresh isolated context — custom tools, permissions, model, memory, and persistent identity · forked subagents available on external builds via CLAUDE_CODE_FORK_SUBAGENT=1 (v2.1.117+) | | ; ⚙️ DEVELOPMENT WORKFLOWS; Others; 🧰 SKILL COLLECTIONS; 💡 TIPS AND TRICKS (82); Tips from Claude code CLI binary; 🎬 VIDEOS / PODCASTS
Prompt / agent / Claude-Code-related paths sampled:
!/claude-jumping.svg
!/claude-speaking.svg
!/tags/claude-certified-architect.svg
!/tags/claude-community-ambassador.svg
!/tags/claude-for-oss.svg
!/tags/claude.svg
!/tags/orchestration-workflow-hd.svg
!/tags/orchestration-workflow.svg
!/thumbnail/daily-workflows.html
!/video-presentation-transcript/1-video-workflow.html
!/video-presentation-transcript/1-video-workflow.md
.claude/agent-memory/weather-agent/MEMORY.md
.claude/agent-memory/weather-agent/readings.md
.claude/agents/development-workflows-research-agent.md
.claude/agents/presentation-claude-gemini.md
.claude/agents/presentation-vibe-coding.md
… plus 64 more
Script-like paths sampled:
.claude/hooks/scripts/hooks.py
.codex/hooks/scripts/hooks.py
Agent/config paths sampled:
.claude/hooks/config/hooks-config.json
.claude/settings.json
.codex/hooks.json
.codex/hooks/config/hooks-config.json
.mcp.json

9. [https://github.com/VoltAgent/awesome-claude-code-subagents VoltAgent/awesome-claude-code-subagents]

Description: A collection of 100+ specialized Claude Code subagents covering a wide range of development use cases
Topics: ai-agent-framework, ai-agent-tools, ai-agents, awesome, awesome-list, claude, claude-ai, claude-code-subagents, claude-subagents, subagents
GitHub signals: 18664 stars, 2131 forks, 11 open issues, default branch main, 179 files in recursive tree.
README intro signals: This repository serves as the definitive collection of Claude Code subagents, specialized AI assitants designed for specific development tasks. bash claude plugin marketplace add VoltAgent/awesome-claude-code-subagents claude plugin install <plugin-name> Examples: bash claude plugin install voltagent-lang # Language specialists claude plugin install voltagent-infra # Infrastructure & DevOps
README headings sampled: Awesome Claude Code Subagents; Installation; As Claude Code Plugin (Recommended); Option 1: Manual Installation; Option 2: Interactive Installer; Option 3: Standalone Installer (no clone required); Option 4: Agent Installer (use Claude Code to install agents); 📚 Categories; 01. Core Development; 02. Language Specialists
Prompt / agent / Claude-Code-related paths sampled:
.github/workflows/enforce-plugin-version-bump.yml
CLAUDE.md
categories/05-data-ai/prompt-engineer.md
categories/06-developer-experience/git-workflow-manager.md
categories/06-developer-experience/mcp-developer.md
categories/09-meta-orchestration/agent-installer.md
categories/09-meta-orchestration/agent-organizer.md
categories/09-meta-orchestration/multi-agent-coordinator.md
categories/09-meta-orchestration/workflow-orchestrator.md
install-agents.sh
Script-like paths sampled:
install-agents.sh
tools/subagent-catalog/config.sh
Agent/config paths sampled:
.claude-plugin/marketplace.json
.claude/settings.local.json
categories/01-core-development/.claude-plugin/plugin.json
categories/02-language-specialists/.claude-plugin/plugin.json
categories/03-infrastructure/.claude-plugin/plugin.json
categories/04-quality-security/.claude-plugin/plugin.json
categories/05-data-ai/.claude-plugin/plugin.json
categories/06-developer-experience/.claude-plugin/plugin.json
categories/07-specialized-domains/.claude-plugin/plugin.json
categories/08-business-product/.claude-plugin/plugin.json
categories/09-meta-orchestration/.claude-plugin/plugin.json
categories/10-research-analysis/.claude-plugin/plugin.json

10. [https://github.com/Piebald-AI/claude-code-system-prompts Piebald-AI/claude-code-system-prompts]

Description: All parts of Claude Code’s system prompt, 24 builtin tool descriptions, sub agent prompts (Plan/Explore/Task), utility prompts (CLAUDE.md, compact, statusline, magic docs, WebFetch, Bash cmd, security review, agent creation). Updated for each Claude Code version.
Topics: claude-code, claude-code-system-prompts, system-prompts
GitHub signals: 9655 stars, 1721 forks, 3 open issues, default branch main, 293 files in recursive tree.
README intro signals: We’ve released Piebald, the ultimate agentic AI developer experience. \ Download it and try it out for free! https://piebald.ai/ > [!important] > NEW (January 23, 2026): We’ve added all of Claude Code’s ~40 system reminders to this list—see System Reminders. This repository contains an up-to-date list of all Claude Code’s various system prompts and their associated token counts as of Claude Code v2.1.122 (April 28th, 2026). It also contains a CHANGELOG.md for the system prompts across 164 versions since v2.0.14. From the team behind Piebald.
README headings sampled: Check out Piebald; Claude Code System Prompts; Extraction; Prompts; Agent Prompts; Sub-agents; Creation Assistants; Slash Commands; Utilities; Data
Prompt / agent / Claude-Code-related paths sampled:
CLAUDE.md
system-prompts/agent-prompt-agent-creation-architect.md
system-prompts/agent-prompt-auto-mode-rule-reviewer.md
system-prompts/agent-prompt-background-agent-state-classifier.md
system-prompts/agent-prompt-bash-command-description-writer.md
system-prompts/agent-prompt-bash-command-prefix-detection.md
system-prompts/agent-prompt-batch-slash-command.md
system-prompts/agent-prompt-claude-guide-agent.md
system-prompts/agent-prompt-claudemd-creation.md
system-prompts/agent-prompt-coding-session-title-generator.md
system-prompts/agent-prompt-conversation-summarization.md
system-prompts/agent-prompt-determine-which-memory-files-to-attach.md
system-prompts/agent-prompt-dream-memory-consolidation.md
system-prompts/agent-prompt-dream-memory-pruning.md
system-prompts/agent-prompt-explore.md
system-prompts/agent-prompt-general-purpose.md
… plus 64 more
Script-like paths sampled:
tools/updatePrompts.js

Reusable research prompt

Research https://share.google/5JRSrbHlPT31c59At. Follow the 10 items named in the article to their GitHub repositories. For each repository, capture the repo URL, GitHub metadata, README-derived purpose, and sampled prompt/agent/config/script paths. Include the exact prompt and the reproducible script used to collect the data in the final wiki result.

Reproducible collection script

The script used to collect the GitHub metadata and file-tree samples is stored locally at /root/workspace/claude_code_repos_research.py. A copy is embedded here for reproducibility:

#!/usr/bin/env python3
"""Research the 10 KDnuggets Claude Code repositories.

Outputs:
  /root/workspace/claude_code_repos_research.json
  /root/workspace/claude_code_repos_research.md

The script uses the public GitHub REST API without authentication. If GITHUB_TOKEN is
set, it will use it to increase rate limits.
"""
from __future__ import annotations

import base64
import json
import os
import re
import textwrap
import time
import urllib.error
import urllib.parse
import urllib.request
from datetime import datetime, timezone
from pathlib import Path
from typing import Any

REPOS = [
    "affaan-m/everything-claude-code",
    "x1xhlol/system-prompts-and-models-of-ai-tools",
    "garrytan/gstack",
    "gsd-build/get-shit-done",
    "shareAI-lab/learn-claude-code",
    "hesreallyhim/awesome-claude-code",
    "davila7/claude-code-templates",
    "shanraisshan/claude-code-best-practice",
    "VoltAgent/awesome-claude-code-subagents",
    "Piebald-AI/claude-code-system-prompts",
]

ARTICLE_URL = "https://www.kdnuggets.com/10-github-repositories-to-master-claude-code"
OUT_JSON = Path("/root/workspace/claude_code_repos_research.json")
OUT_MD = Path("/root/workspace/claude_code_repos_research.md")

TOKEN = os.environ.get("GITHUB_TOKEN")
HEADERS = {
    "Accept": "application/vnd.github+json",
    "X-GitHub-Api-Version": "2022-11-28",
    "User-Agent": "HermesResearchBot/1.0",
}
if TOKEN:
    HEADERS["Authorization"] = f"Bearer {TOKEN}"

def gh(path_or_url: str, *, raw: bool = False, retries: int = 3) -> Any:
    url = path_or_url if path_or_url.startswith("http") else f"https://api.github.com{path_or_url}"
    headers = dict(HEADERS)
    if raw:
        headers["Accept"] = "application/vnd.github.raw"
    last_exc: Exception | None = None
    for attempt in range(retries):
        req = urllib.request.Request(url, headers=headers)
        try:
            with urllib.request.urlopen(req, timeout=45) as resp:
                body = resp.read()
                ctype = resp.headers.get("content-type", "")
                if raw or "json" not in ctype:
                    return body.decode("utf-8", errors="replace")
                return json.loads(body.decode("utf-8"))
        except urllib.error.HTTPError as e:
            last_exc = e
            if e.code in (403, 429, 500, 502, 503, 504):
                time.sleep(2 ** attempt)
                continue
            raise
        except Exception as e:
            last_exc = e
            time.sleep(2 ** attempt)
    raise RuntimeError(f"Failed {url}: {last_exc}")

def get_readme(repo: str) -> str:
    try:
        data = gh(f"/repos/{repo}/readme")
        if isinstance(data, dict) and data.get("content"):
            return base64.b64decode(data["content"]).decode("utf-8", errors="replace")
    except Exception as e:
        return f"[README unavailable: {e}]"
    return ""

def classify_paths(paths: list[str]) -> dict[str, list[str]]:
    prompt_patterns = re.compile(r"(^|/|[-_])(prompt|prompts|system[-_ ]?prompt|claude|command|commands|agent|agents|skill|skills|workflow|workflows|mcp|hook|hooks)(/|[-_.]|$)", re.I)
    script_ext = re.compile(r"\.(sh|bash|zsh|fish|py|js|mjs|cjs|ts|tsx|jsx|rb|go|rs|php|pl|lua|ps1)$", re.I)
    config_ext = re.compile(r"\.(ya?ml|json|toml|ini|env|mdc)$", re.I)

    prompt_like = []
    script_like = []
    configs = []
    docs = []
    for p in paths:
        lower = p.lower()
        if prompt_patterns.search(p) or lower in {"claude.md", ".claude/settings.json"}:
            prompt_like.append(p)
        if script_ext.search(p) or "/scripts/" in lower or lower.startswith("scripts/") or "/bin/" in lower:
            script_like.append(p)
        if config_ext.search(p) and any(tok in lower for tok in ["claude", "mcp", "agent", "hook", "command", "skill"]):
            configs.append(p)
        if lower.endswith((".md", ".mdx", ".txt")):
            docs.append(p)
    return {
        "prompt_like": sorted(set(prompt_like))[:80],
        "script_like": sorted(set(script_like))[:80],
        "agent_configs": sorted(set(configs))[:80],
        "docs": sorted(set(docs))[:80],
    }

def extract_readme_signals(readme: str) -> dict[str, Any]:
    lines = [line.rstrip() for line in readme.splitlines()]
    title = ""
    for line in lines[:40]:
        if line.startswith("#"):
            title = line.lstrip("# ").strip()
            break
    headings = [re.sub(r"^#+\s*", "", l).strip() for l in lines if l.startswith("#")][:20]
    bullets = []
    for l in lines:
        s = l.strip()
        if re.match(r"^[-*+]\s+", s) and 20 <= len(s) <= 180:
            bullets.append(re.sub(r"^[-*+]\s+", "", s))
        if len(bullets) >= 12:
            break
    # Compact human-readable excerpt from first non-badge paragraphs.
    paras = []
    cur = []
    for line in lines[:200]:
        s = line.strip()
        if not s:
            if cur:
                para = " ".join(cur)
                if len(para) > 60 and "img.shields.io" not in para and not para.startswith("["):
                    paras.append(para)
                cur = []
            continue
        if s.startswith("#") or s.startswith("!") or s.startswith("<"):
            continue
        cur.append(s)
        if len(paras) >= 3:
            break
    return {"title": title, "headings": headings, "sample_bullets": bullets, "intro": paras[:3]}

def fetch_repo(repo: str) -> dict[str, Any]:
    meta = gh(f"/repos/{repo}")
    readme = get_readme(repo)
    tree_items = []
    try:
        tree = gh(f"/repos/{repo}/git/trees/{urllib.parse.quote(meta['default_branch'])}?recursive=1")
        tree_items = [item for item in tree.get("tree", []) if item.get("type") == "blob"]
    except Exception as e:
        tree_items = [{"path": f"[tree unavailable: {e}]", "type": "error"}]
    paths = [i.get("path", "") for i in tree_items]
    classes = classify_paths(paths)
    return {
        "full_name": repo,
        "url": meta.get("html_url"),
        "description": meta.get("description"),
        "stars": meta.get("stargazers_count"),
        "forks": meta.get("forks_count"),
        "watchers": meta.get("subscribers_count"),
        "open_issues": meta.get("open_issues_count"),
        "license": (meta.get("license") or {}).get("spdx_id"),
        "topics": meta.get("topics", []),
        "default_branch": meta.get("default_branch"),
        "created_at": meta.get("created_at"),
        "pushed_at": meta.get("pushed_at"),
        "updated_at": meta.get("updated_at"),
        "homepage": meta.get("homepage"),
        "file_count": len(paths),
        "readme_signals": extract_readme_signals(readme),
        "paths": classes,
    }

def fmt_list(items: list[str], limit: int = 12) -> str:
    if not items:
        return "None found in repository tree."
    out = "\n".join(f"* `{p}`" for p in items[:limit])
    if len(items) > limit:
        out += f"\n* ... plus {len(items) - limit} more"
    return out

def build_markdown(data: list[dict[str, Any]]) -> str:
    now = datetime.now(timezone.utc).strftime("%Y-%m-%d %H:%M UTC")
    lines = []
    lines.append("= 10 Claude Code GitHub Repositories =")
    lines.append("")
    lines.append(f"Research date: {now}")
    lines.append(f"Source article: [{ARTICLE_URL} KDnuggets: 10 GitHub Repositories To Master Claude Code]")
    lines.append("")
    lines.append("== Summary ==")
    lines.append("This page follows the 10 repositories named in the KDnuggets article and records what each repo is for, how active/popular it appears from GitHub metadata, and where prompts, Claude Code configuration, agents, skills, hooks, or scripts appear in the repository tree.")
    lines.append("")
    lines.append("{| class=\"wikitable sortable\"")
    lines.append("! # !! Repository !! Stars !! Forks !! License !! Last pushed !! Main value")
    for i, r in enumerate(data, 1):
        desc = (r.get("description") or "").replace("|", "&#124;")
        lines.append("|-")
        lines.append(f"| {i} || [{r['url']} {r['full_name']}] || {r.get('stars')} || {r.get('forks')} || {r.get('license') or '—'} || {str(r.get('pushed_at',''))[:10]} || {desc}")
    lines.append("|}")
    lines.append("")
    lines.append("== Repository notes ==")
    for i, r in enumerate(data, 1):
        sig = r.get("readme_signals", {})
        lines.append(f"=== {i}. [{r['url']} {r['full_name']}] ===")
        lines.append(f"* **Description:** {r.get('description') or 'No GitHub description.'}")
        lines.append(f"* **Topics:** {', '.join(r.get('topics') or []) or 'None listed'}")
        lines.append(f"* **GitHub signals:** {r.get('stars')} stars, {r.get('forks')} forks, {r.get('open_issues')} open issues, default branch `{r.get('default_branch')}`, {r.get('file_count')} files in recursive tree.")
        if sig.get("intro"):
            lines.append("* **README intro signals:** " + " ".join(sig["intro"])[:900])
        if sig.get("headings"):
            lines.append("* **README headings sampled:** " + "; ".join(sig["headings"][:10]))
        lines.append("* **Prompt / agent / Claude-Code-related paths sampled:**")
        lines.append(fmt_list(r["paths"].get("prompt_like", []), 16))
        lines.append("* **Script-like paths sampled:**")
        lines.append(fmt_list(r["paths"].get("script_like", []), 12))
        if r["paths"].get("agent_configs"):
            lines.append("* **Agent/config paths sampled:**")
            lines.append(fmt_list(r["paths"].get("agent_configs", []), 12))
        lines.append("")
    lines.append("== Reusable research prompt ==")
    lines.append("<pre>")
    lines.append("Research https://share.google/5JRSrbHlPT31c59At. Follow the 10 items named in the article to their GitHub repositories. For each repository, capture the repo URL, GitHub metadata, README-derived purpose, and sampled prompt/agent/config/script paths. Include the exact prompt and the reproducible script used to collect the data in the final wiki result.")
    lines.append("&lt;/pre&gt;")
    lines.append("")
    lines.append("== Reproducible collection script ==")
    lines.append("The script used to collect the GitHub metadata and file-tree samples is stored locally at `/root/workspace/claude_code_repos_research.py`. A copy is embedded here for reproducibility:")
    lines.append("<pre>")
    script = Path(__file__).read_text()
    # Avoid nested &lt;/pre&gt; breakage if the script is edited later.
    lines.append(script.replace("&lt;/pre&gt;", "&lt;/pre&gt;"))
    lines.append("&lt;/pre&gt;")
    lines.append("")
    lines.append("")
    lines.append("")
    return "\n".join(lines)

def main() -> None:
    data = []
    for repo in REPOS:
        print(f"Fetching {repo}...", flush=True)
        try:
            data.append(fetch_repo(repo))
        except Exception as e:
            data.append({"full_name": repo, "url": f"https://github.com/{repo}", "error": str(e), "paths": {"prompt_like": [], "script_like": [], "agent_configs": []}})
    OUT_JSON.write_text(json.dumps({"article_url": ARTICLE_URL, "generated_at": datetime.now(timezone.utc).isoformat(), "repos": data}, indent=2), encoding="utf-8")
    OUT_MD.write_text(build_markdown(data), encoding="utf-8")
    print(f"Wrote {OUT_JSON}")
    print(f"Wrote {OUT_MD}")

if __name__ == "__main__":
    main()