A curated collection of Claude Code projects, agent workspace blueprints, and related resources — organized by use case. Most patterns here adapt to other agentic AI CLIs and frameworks.
Browse online · Plugins Marketplace · Documentation portal · What are Claude Spaces?
🧭 Start here: Marketplace Guide. This is the suggested way into my Claude Code Plugins Marketplace. Install it (
/plugin install marketplace-guide@danielrosehill) and ask Claude "what's in the marketplace?" or "is there a plugin for X?". It browses every plugin by theme (Israel & Hebrew, homelab, Linux desktop, Claude Code tooling, shopping, writing…), recommends plugins for a task and shows what's new, all read live from the marketplace manifest.
Workspaces by Domain
- Systems Administration
- Productivity & Planning · Legal · Health & Wellbeing · Communications & Writing · Financial Planning · Career · Business · Privacy & Anonymity · Technology & Hardware · Marketing
- Research
- Argument and Perspective Exploration
Configuration & Tooling
Extensions & Scaffolds
- Plugins
- Templates / Scaffolds — recommended way to spin up a new workspace
- Slash Commands
Other
I've been using Claude Code daily for about six months — for development, but also audio editing, legal research, SEO analysis, health documentation, systems administration, and a long tail of non-code use cases. This index is the result: a collection of agent workspaces (repositories structured as self-contained environments for a specific activity) alongside supporting tooling — plugins, context files, MCP servers, and slash commands.
If there's a common thread, it's treating Claude Code less as a coding assistant and more as a general-purpose agent workspace that happens to run in a terminal.
More context: the Agent Workspace Model, growth chart, praise
All workspaces in this index follow the same pattern: a Git repository isn't just for code — it can serve as a complete, self-contained workspace for any activity. Each workspace uses a defined folder structure, a CLAUDE.md for agent instructions, slash commands, MCP configurations, and subagent definitions to create a purpose-built environment.
This pattern has been applied to everything from sysadmin and remote server management to legal research, health documentation, and financial planning — domains that have nothing to do with software development.
Primitives globally, scaffolds per-cluster. The tooling has been consolidated into 28 cluster plugins (see Plugins) — each one ships the domain primitives globally (commands, skills, agents for that cluster) and provisions a project scaffold on demand. So rather than forking a separate template repo for each new workflow, you install the relevant cluster plugin once and ask Claude Code to provision a scaffold wherever you need one.
"This is either the work of a prolific genius, or a very clever bot (or both), although it hardly matters because the quality is so good - an index of 75+ Claude Code repositories published by the author... CMS, system design, deep research, IoT, agentic workflows, server management, personal health... If you spot the lie, let me know, otherwise please check these out."
For the record: I'm a real human (danielrosehill.com). The repos and workspaces in this index are generated with Claude Code but human-designed and refined.
Projects involving using Claude for local or remote systems administration as distinct from development-related projects.
See also: The Claude Code Sysadmin Workspaces Index is a dedicated sub-index for all sysadmin workspace templates.
Workspace for managing bash aliases with YADM synchronization support.
Collection of bash aliases for common Claude Code operations on Linux.
GUI concept for Claude-driven LAN management, covering device discovery and network operations.
Planning workspace for managing a Linux desktop with an agentic CLI, running raw notes through a refine/analyse/workflow pipeline.
Multi-plugin installation and setup utility for orchestrating everyday Linux desktop use through Claude Code.
Template using Claude Code as an Ansible stand-in for repetitive environment setup and supervision.
Concept for deploying Claude Code into recovery shell environments for AI-assisted system repair.
Raspberry Pi jump host running the Claude Code CLI over Cloudflare Access and Tailscale as an emergency recovery surface for a Linux workstation.
Custom GRUB boot entry integrating Claude CLI into Linux system recovery workflows.
Projects specifically targeting KDE Plasma desktop integration and Linux desktop workflows with Claude Code.
KDE Dolphin right-click context menu actions (service menus) for launching Claude Code in various Konsole window layouts, including single terminal, split panes, and multi-instance grids.
Launch utility for pairing Claude Code with Konsole on KDE Plasma.
Early standalone Claude Code plugin with slash commands for managing Conda environments and packages.
Per-machine, local-first SQLite log of everything the agent installed, fixed or diagnosed on a Linux desktop, so later sessions can query history before acting. Pairs with linux-desktop-plugin.
KDE Dolphin right-click service menus that open the selected directory in Claude Code or Codex, in normal or permissive mode.
Plan for a KDE-native runner that starts and supervises Claude Code sessions from the desktop.
Claude Code plugin that turns a LAN into working memory: devices, services, operating systems, ports and topology captured in a greppable knowledge store that grows as you and Claude discover things.
Early standalone Claude Code plugin with slash commands for server monitoring, deployment and administration.
Claude Code plugin for administering a Synology NAS over SSH, with guided first-run setup, persistent context and commands for shares, volumes, mounts and storage monitoring.
Workspaces for decision-making, personal planning, file management, and general-purpose productivity workflows.
See also: The Budgeting, Personal Planning, Career, Purchasing, Shopping, and Ideation & Planning cluster plugins in the Plugins section cover these domains.
Index of templates, skills and demos for agent-driven purchase research, built on decoupling the requirements spec from the current market.
Staged, agent-supported process for evaluating any purchase over about $50: capture the spec, find the requirements, survey the market, then score candidates against the spec.
Codifies the iterative spec-refinement loop used with Claude for builds, hardware purchases and home-lab decisions — draft spec, correction, refinement — as reusable skills.
Model workspace for a Claude shopping research tool customised by shop whitelist and task, with input and output folders.
Demo of Claude Code as a consumer purchasing assistant: a written spec plus retailer catalogue screenshots go in, a ranked recommendation report comes out (worked example: a portable Bluetooth speaker).
Work-in-progress CLI and workspace for importing recordings from a digital voice recorder, organising them into dated folders, classifying them with AI and transcribing them.
Workspaces and templates for legal research, case management, and evidence handling workflows.
See also: The Legal & Investigative cluster plugin in the Plugins section covers this domain — evidence logging, document analysis, redaction, and brief generation.
Claude Code template for building version-controlled containers for legal files and case material.
Workspaces and templates for health documentation, medical visit management, therapy tracking, and health-related research.
See also: The Personal Planning cluster plugin in the Plugins section covers this domain — diary, health, therapy, preparedness, and personal development variants.
Research notebook investigating ADHD medication access, structured for sustained Claude-assisted enquiry.
Subindex of Claude Code templates, plugins and workspaces for mental health work a person does on their own material: organising events, tracking therapy, preparing for appointments.
GitHub template for working through one bounded personal issue with Claude Code — a career situation, a relationship, a bereavement — with sessions, threads and timelines kept in one workspace per issue.
Workspaces and templates for content creation, blog management, writing workflows, and communications strategy.
Pattern spec for writing full-length reference and instructional books with Claude Code, using teams of subagents and the repository as shared memory, through to a print-on-demand paperback.
Per-project workspace skeleton for producing one AI-generated video; the state half of a pair whose operational half is the ai-video-producer plugin (scripting, image-to-video and lip-sync pipelines).
Workspace scaffold implementing the Agentic Book Writing pattern: a book.toml manifest, spec and interview folders, binding prose and graphics style guides, and a status script for subagent-driven drafting.
Automated workflow for sending polished update emails about website changes via Resend MCP.
Personalized text rewriting rules that consolidate into a slash command for refining AI-generated documentation.
Instruction-only agent skills that keep Hebrew, Arabic, Farsi and Urdu text right-to-left on every surface an agent writes to (Gmail, Resend, Google Docs/Sheets/Slides) and verify the setting took. Works in Claude and ChatGPT.
Early standalone Claude Code plugin with slash commands for proofreading, style editing and text transformation.
Workspaces and templates for budgeting, purchasing decisions, and personal finance management.
See also: The Budgeting, Purchasing, and Shopping cluster plugins in the Plugins section cover these domains.
Claude Code workspace for FX pair analysis — historical data, quant and vol modelling, geopolitical context, and Typst PDF reports.
Work-in-progress skills basket for buyer-side Israeli residential mortgage research — track mix, Bank of Israel rules, DSR/LTV limits and bank quirks — before verified skills are folded into the Israel agent-skills plugin.
Work-in-progress skills basket for Israeli property due diligence (Tabu extracts, building files, usage permits, comparables), aimed at non-professional buyers and tenants.
Workspaces and templates for job searching, career planning, and professional development.
See also: The Career cluster plugin in the Plugins section covers this domain — role logging, offer comparison, application tracking, and salary benchmarking.
Workspaces and templates for business planning, idea evaluation, and organizational continuity.
See also: The Ideation & Planning cluster plugin in the Plugins section covers business idea evaluation, decision frameworks, and simulation workflows.
Workspaces and templates for document redaction, identity protection, and PII obfuscation.
See also: The Legal & Investigative cluster plugin in the Plugins section includes redaction and document-obfuscation workflows. For broader system hardening see the Security Checkup plugin.
Assistant configuration for redacting personally identifiable details from documents.
Workspaces for hardware planning, PC builds, and technology procurement.
See also: The Purchasing (includes a tech-procurement variant), Sysadmin & Homelab, and HP5200 Printer plugins in the Plugins section cover these domains.
Dedicated macropad for driving Claude Code — approve, stop, new session, transcript navigation — with a host-driven status LED that lights when Claude is waiting on you.
Ergonomic and geometric research into a display for running a grid of at least eight concurrent Claude Code sessions, measured in text cells rather than inches, plus the software layer for navigating the grid.
Non-code planning workspace for a home media setup: ideas, voice notes and reference material refined into requirements and decisions.
Workspaces for SEO, web analytics, PR monitoring, and media tracking.
See also: The PR & Media Work cluster plugin in the Plugins section covers coverage scanning, press summarisation, response drafting, and comms strategy.
Projects using Claude and agentic systems for deep research, report generation, and information synthesis.
See also: The Research Space cluster plugin in the Plugins section covers deep research, technical research, OSINT, geo-reaction, stack, ecosystem, and competitor research workflows.
See full list in the dedicated research page → (12 entries)
Projects using AI for synthesized debate to explore various perspectives, including policy modeling and analysis.
Simulated conference in which agentic personas (HR leads, CEOs, doctors, artists, sceptics) each deliver a three-minute talk on "AI in 2026", rendered with TTS.
Workspace for challenging personal beliefs through AI-generated counterarguments and rebuttals.
Exploratory model running multiple agents as a panel debate.
Claude Code workspace for finding people who hold specific viewpoints within a policy debate, surfacing voices that keyword search misses.
Projects exploring using Claude and related tooling for personalized user engagement, including through RAG, interviewing methods, and context injection.
GitHub template repository pre-seeded with every AI Dev Workspace Primitive as a stub, with instructions for the user or agent to fill each one in.
October 2025 template for AI-assisted development with a layered CLAUDE.md hierarchy that guides the agent through project stages.
Index of root-level UPPERCASE.md context primitives for AI-assisted repos (SPEC.md, RULES.md, TOOLS.md, SECRETS.md, GROUNDING.md and more), each published as its own repo and injectable via CLAUDE.md/AGENTS.md.
Automation workspace for batch-adding CLAUDE.md files across multiple GitHub repositories.
Pre-configured CLAUDE.md templates for managing different repository types.
Redacted point-in-time dump of /context from a heavily-pluginned Claude Code session, analysing where the context budget actually goes and what is eagerly versus lazily loaded.
Index of the user-level layer of a Claude Code setup, treated as one system — entrypoint, reference context, habits and discovery — covering what loads before any task is known.
Prompt template for verifying the correct Claude model variant at conversation start.
Depersonalised reference pattern for organising user-level ~/.claude/ context: a lean top-level CLAUDE.md that routes to topical context/*.md files (system environment, git rules, MCP usage, file organisation, etc.) loaded only when relevant.
Pattern for maintaining separate CLAUDE.md files for developers and for end users of the same repository.
Workflow methodology for separating human-authored context from structured AI agent briefings.
Splits a user-level system prompt into one file per standing habit, assembled into a pasteable block and a JSON index. Ships skills to install a subset into any CLAUDE.md and to reconcile drift when the prompt and the repo disagree.
Working CLAUDE.md for a home directory on an Ubuntu Linux desktop.
Automatically generates contextual CLAUDE.md files across a Linux desktop filesystem.
Annotated corpus of writing samples used as ground truth for a personal writing style, with reusable methodology for building a tone-of-voice reference for AI agents.
Tools for managing public and private CLAUDE.md files with security-focused git configuration.
Pattern for splitting a home-level CLAUDE.md into directives plus on-demand context files, optimised for context handling.
Q&A notes on where context bloat accrues given the current shape of the Claude Code harness and its primitives.
CLAUDE.md snippets and slash commands telling Claude to silently infer around transcription errors from voice typing and stray keystrokes from one-handed or distracted typing.
Components and tooling for multi-agent development and orchestration frameworks.
Design model for an orchestration agent that buffers a human from a batch of working sub-agents — one consolidated brief per cycle, a budgeted override channel, and resumable approval gates.
Pattern for handing tasks between code-generation agents working on different machines through shared handover docs, giving supervising humans an audit trail of what each agent did.
MCP server enabling encrypted peer-to-peer communication between Claude Code instances on localhost or LAN.
Framework for assembling context-optimized multi-agent crews with minimal overlap.
Launchpad for creating standardized workspace templates conforming to the Agent Workspace Model v1.1 spec, with slash commands to generate, validate, and publish new workspaces.
Workspace model for one operator running agentic AI across many repositories all day, borrowed from air traffic control: a voice-first shift manager routes work across concurrent Claude Code sessions. Published as a 26-page typeset document.
Specification for a local proxy that assigns priorities to concurrent Claude Code sessions and shapes their aggregate traffic, so rate limiting degrades background work instead of landing at random. Includes a prior-art survey of the multi-agent proxy and queueing landscape.
Curated toolkit of 74+ Claude Code configurations for development workflows and multi-agent coordination.
Collection of system prompts and configurations for development, operational, and creative tasks.
Curated collection of Claude Code projects and resources organized by category.
Architecture for Claude Code plugins that do sustained work: a public plugin, a private workspace instantiated from a template repo, and a vault at ~/.claude-plugins/<name>/ for user data.
Multimodal workspace for defining Claude Code skills from voice notes, videos, images and text, or written specs.
Idea capture for a meta-plugin holding a personal vault of plugins and MCP servers, activated selectively per project to mitigate eager plugin description loading at user level.
Interactive tool for discovering, selecting, and cloning Claude Workspace templates.
Projects related to Claude and MCP tooling and setup.
Early standalone Claude Code plugin with slash commands for AI tool management and MCP server configuration.
Generator for creating MCP server configuration commands for Claude Code.
Curated index of MCP servers organized into 14+ categories for extending Claude Code.
Home-folder CLAUDE.md snippet giving an agent guidance on which MCP server to reach for.
Template /install-mcp slash command that routes MCP server installations across multiple MetaMCP instances using a tiered preference system.
Guide for instructing AI agents on how to provision and manage MCP server connections according to user-specific preferences, with a tiered decision matrix.
Documentation for integrating Claude Code with MCPM external MCP server manager.
Curated collection of 35+ MCP servers with interactive installer across 15+ categories.
All plugins registered in the danielrosehill marketplace. Install any of these with /plugin install <name>@danielrosehill.
See full list in the dedicated plugins page → (179 entries)
Scaffolds used to be distributed as ~100 standalone template repos and a New-Repo-From-Template plugin. That pattern was retired in the April 2026 reshape.
Scaffolds now live inside the cluster plugins. Each of the 28 cluster plugins bundles the workspace primitives for its domain (commands, skills, agents, MCP configs) and exposes a provisioning skill that writes a fresh per-project scaffold on demand — so instead of cloning a template repo, you install the relevant plugin once and ask Claude Code to scaffold a new workspace for whatever project you're starting.
See the Plugins section above for the full cluster list.
Individual slash commands, sometimes integrated into other plugins or sometimes just for use at the user level.
See also: The Claude Slash Commands repo serves as both a 350+ command library and the dedicated index for all slash command repos.
Adds AI/human attribution sections to README files for transparent tool usage documentation.
Slash command deploying a sub-agent for codebase optimization and dead code removal.
Flips Claude Code into autonomous, action-first mode — ships a per-session /calls-the-shots slash command plus an optional always-on snippet injected into ~/.claude/CLAUDE.md.
System administration slash commands for Linux desktop environments.
Slash command that condenses bloated CLAUDE.md files to essentials and organizes supplementary context into a structured agent-context/ folder.
General-purpose slash command library for various Claude Code workflows.
Drop-in system prompt and slash command that make coding agents capture what they discover — undocumented APIs, auth flows, dead ends — instead of losing it when the session ends.
Early standalone Claude Code plugin with slash commands and agents for code generation, refactoring and general development.
Early standalone Claude Code plugin with slash commands and agents for Git version control and GitHub repository management.
Slash commands encouraging use of existing libraries instead of building custom solutions.
Slash command that prunes a repo's CLAUDE.md to essentials and scaffolds a topic-based context/ store, an ingestion pipeline for raw materials, and subagents for context maintenance.
Other projects including meta-resources, feedback, and utilities that span multiple categories.
See full list in the dedicated misc page → (32 entries)
Early-stage repositories: ideas, planning notes, placeholders and work-in-progress skills baskets that have not yet become a working tool, workspace or plugin. Some are stubs. They are listed so the idea is findable, not because there is anything to install.
Idea: a per-turn router that sends each message to the current task, an existing parallel worker or a new session, keeping unrelated tasks out of one agent's context.
Placeholder for planning a Claude Code background hook.
Empty placeholder for experiments with CLAUDE.md structure.
Stub for planning notes on the plugins and skills mechanism (May 2026).
Planning repo for Claude placing a real outbound phone call to deliver a one-time, task-specific spoken message on the user's behalf.
Stub workspace for agent skills that check medication availability in Israel for patients, to be folded into the Israel agent-skills plugin.
Planned multi-label classifier that routes voice-note transcripts to task-specific subagents (email, calendar, shopping list, to-do, dev spec), each holding one MCP tool.