Skip to content

Knowledge Worker Walkthrough

This walkthrough is for you if you use Claude for research, writing, analysis, or content creation, and want to bring your personal context and preferred tools into every conversation. You'll discover pre-built bundles on the marketplace, deploy them with one click, connect external tools via MCP servers, and teach Claude about your domain so it remembers you across sessions.

No terminal required. Everything happens in the web UI.

Who This Is For

You are a knowledge worker if you:

  • Use Claude daily for writing, research, analysis, or domain-specific tasks
  • Want Claude to remember your preferences, terminology, and past work
  • Use external tools (databases, APIs, document systems) that Claude should access
  • Prefer visual interfaces to command-line tools
  • Want production-ready artifacts without building everything yourself

Time to complete: ~20 minutes
Complexity: Beginner (web UI only)

Prerequisites

What you need

  • SkillMeat installed — See Quickstart Guide for one-line install (pip install skillmeat)
  • SkillMeat running — Start with skillmeat web dev and open http://localhost:3000
  • Claude Code or Claude Desktop — Set up and ready to use
  • Internet access — To browse and download from the marketplace
  • A Claude Code project (optional) — You can create one as you go

Step 1: Open the Marketplace

Start the SkillMeat web UI:

skillmeat web dev

Wait for both servers to start, then open http://localhost:3000 in your browser.

Browse by Use Case

  1. In the left sidebar, click Marketplace
  2. You'll see curated bundles organized by domain:
  3. Writing & Content — Skills for drafting, editing, structuring documents
  4. Research & Analysis — Tools for literature review, data synthesis, visualization
  5. Knowledge Management — Context capture, memory systems, note organization

  6. Integration Suites — Pre-configured tool bundles for specific workflows

  7. Read the bundle description — Each shows:

  8. What skills it includes
  9. Who it's designed for
  10. Prerequisites (e.g., API keys for external tools)
  11. What you'll be able to do after installing

CLI alternative

Prefer the command line? Use skillmeat marketplace search --type bundle --query "writing" to search by keyword.

Step 2: Install a Curated Bundle

Pick a bundle that matches your workflow. Let's say you choose "Research Assistant Bundle" (includes research note skills, citation management, and synthesis tools).

One-Click Install

  1. Click the bundle card
  2. Read the Security & Permissions notice (summarizes what the bundle can access)
  3. Click Install to Collection
  4. Wait ~10 seconds — you'll see "Installation complete"

Your bundle is now in your personal collection.

What happened

SkillMeat downloaded the bundle from the marketplace, verified its authenticity, and stored it in your collection. No manual steps. No configuration required yet.

CLI alternative

In terminal: skillmeat add bundle anthropics/bundles/research-assistant

Step 3: Deploy to Your Claude Code Project

Now make the bundle available in Claude Code.

Deploy via Web UI

  1. Click Projects in the left sidebar
  2. Select your project (or create a new one by clicking New Project → name it, e.g., "My Research")
  3. Click the Artifacts tab
  4. Click Deploy and search for your bundle name
  5. Check the items you want to deploy (usually all by default)
  6. Click Deploy

You'll see a progress bar. When it completes, your artifacts are now in your project's .claude/ directory.

CLI alternative

In terminal, navigate to your project and run: 

skillmeat deploy research-assistant

Verify Deployment

Back in the Web UI:

  1. Refresh the Projects tab
  2. Select your project and click Deployed
  3. You should see your bundle's artifacts listed with version info and deployment timestamps

Your Claude Code project now has the research tools ready to use.

Step 4: Set Up External Tool Access (MCP Servers)

Some bundles include MCP servers — these let Claude access databases, APIs, and other external tools. You configure them in the web UI without touching configuration files.

Add an MCP Server

  1. In your project view, click MCP Servers tab (if visible)
  2. Click Add Server
  3. Select the server type (e.g., "PostgreSQL", "File System", "Web Search")
  4. Enter connection details:
  5. For databases: host, port, credentials (securely stored)
  6. For APIs: API key or auth token

  7. For file systems: path or network mount

  8. Click Test Connection to verify it works

  9. Click Save

The MCP server is now active. When you use Claude Code with this project, Claude can query your database or access your external tools.

Credentials Security

Credentials are encrypted at rest. Never commit them to Git. SkillMeat stores them in your local collection, not in version-controlled files.

CLI alternative & Advanced Setup

For multi-server configurations or advanced options, see MCP Management Guide.

Step 5: Capture Personal Context (Memory)

Claude can remember your preferences and past work within a session. With the Memory System, you teach it once and it recalls across sessions.

What to Capture

Think about:

  • Your writing style — "I prefer short paragraphs with one main idea each"
  • Key terminology — Domain-specific terms, acronyms, preferred definitions
  • Project context — What you're researching, the timeline, stakeholders
  • Past decisions — Why you chose certain approaches (so Claude doesn't suggest the same again)
  • Tool preferences — Which external tools you use and when

Add a Memory Item

  1. In your project, click Memory tab
  2. Click Add Memory
  3. Write a clear, concise note (1–3 sentences):
  4. Example: "I use the 'STAR framework' (Situation, Task, Action, Result) for narrative analysis. Always structure findings this way unless told otherwise."
  5. Example: "I prefer Chicago Manual of Style citations over APA."

  6. Example: "Our research focuses on climate adaptation in Southeast Asia, 2020–2030 timeline."

  7. Optionally assign a category (style, domain-knowledge, project-context, preferences)

  8. Click Save

Claude will see this memory when you start a conversation in Claude Code with this project.

Why This Matters

Instead of repeating "use STAR framework" every session, Claude sees your memory and adapts automatically. You save time and get more consistent outputs.

Advanced Memory Workflows

For deeper context packs and advanced memory features, see Memory System Guide.

Step 6: Verify Everything Works

Checklist

  • [ ] Marketplace: You browsed bundles and installed one
  • [ ] Deployed artifacts: Artifacts appear in your project's Deployed tab
  • [ ] MCP servers (if applicable): You configured external tool access and tested it
  • [ ] Memory items: You added at least one memory note about your domain or style

Use It in Claude Code

  1. Open Claude Code or Claude Desktop
  2. Load your project (open the folder or select it in the UI)
  3. In a conversation, reference your research:
  4. "Help me structure this analysis using the STAR framework"
  5. "Search the database for climate studies since 2020"
  6. "Draft a research proposal following Chicago style"

Claude has access to your artifacts, external tools, and personal memory. It adapts to your style and context automatically.

Troubleshooting

"Bundle won't install"

  1. Check your internet connection
  2. Verify the bundle is compatible with your SkillMeat version:
  3. In Marketplace, click the bundle → check "Requires SkillMeat v0.30+"
  4. In terminal: skillmeat --version
  5. If bundle is very new, you may need to update SkillMeat: 
    pip install --upgrade skillmeat

"MCP server test fails"

  • Database: Check host, port, and credentials. Verify the database is running.
  • API: Verify your API key is correct and hasn't expired.
  • File system: Check the path exists and you have read permissions.

"Memory items don't appear in Claude"

  1. Make sure you've deployed artifacts to the project
  2. In Claude Code, confirm the project folder is open
  3. Memory appears in new conversations, not retroactively in old ones
  4. Check the Memory tab → your items should show status "Active"

What Just Happened

You've successfully:

  1. ✅ Discovered and installed a pre-built bundle from the marketplace
  2. ✅ Deployed artifacts to your Claude Code project (no CLI needed)
  3. ✅ Configured MCP servers to give Claude access to external tools
  4. ✅ Captured personal context so Claude remembers your preferences across sessions

Your knowledge work is now enhanced: Claude has your tools, understands your domain, and remembers how you work.

Next Steps

Now that you have artifacts and memory in place, explore deeper workflows:

Getting Help

  • Web UI help: Hover over any icon or label for tooltips
  • Marketplace questions: Click a bundle → scroll to Support section
  • Report issues: File an issue on GitHub

Ready to go deeper? Check out Marketplace Search & Install Guide for advanced search and lifecycle management.

Time to complete: ~20 minutes
Complexity: Beginner
Prerequisites met: ✅