diff --git a/.claude/commands/setup-firecrawl.md b/.claude/commands/setup-firecrawl.md
new file mode 100644
index 00000000..73fe505f
--- /dev/null
+++ b/.claude/commands/setup-firecrawl.md
@@ -0,0 +1,305 @@
+---
+name: setup-firecrawl
+description: Interactive setup wizard for Firecrawl web scraping scripts
+allowed-tools: [Read, Write, Bash, AskUserQuestion]
+argument-hint:
+  "(optional) skip steps with 'quick' if you already have API key configured"
+---
+
+# Firecrawl Setup Wizard
+
+This command guides you through setting up Firecrawl for web scraping, enabling
+you to save articles, documentation, and web content directly to your vault as
+clean, searchable markdown.
+
+## Task
+
+Set up Firecrawl web scraping by:
+
+1. Explaining the benefits of Firecrawl for research workflows
+2. Getting a Firecrawl API key (300 free credits to start)
+3. Configuring the FIRECRAWL_API_KEY environment variable
+4. Understanding the available scripts (single and batch scraping)
+5. Testing with a sample scrape
+6. Showing example research workflows
+
+## Process
+
+### 1. Welcome & Explain Benefits
+
+Start by explaining what Firecrawl enables:
+
+**The Research Game-Changer:** Instead of Claude reading webpages and
+summarizing (losing details), Firecrawl saves the FULL text as markdown in your
+vault. This means:
+
+- **Permanent archive**: Articles stay in your vault forever
+- **Full context**: Complete text, not summaries
+- **Searchable**: Find anything across thousands of saved articles
+- **No limits**: Build a massive research library without context limits
+- **Clean format**: Beautiful markdown, not messy HTML
+
+**Perfect for:**
+
+- Research projects (save all related articles)
+- Documentation archives (keep important guides)
+- Knowledge bases (build your own reference library)
+- Reading lists (save now, read later)
+- Academic research (preserve sources)
+
+### 2. Prerequisites Check
+
+Verify required tools are installed:
+
+```bash
+curl --version  # Should be installed
+jq --version    # Should be installed (for JSON parsing)
+```
+
+If jq is missing:
+
+- **macOS**: `brew install jq`
+- **Linux**: `apt-get install jq` or `yum install jq`
+- **Windows**: Use WSL or Git Bash
+
+### 3. Get Firecrawl API Key
+
+Ask: "Do you already have a Firecrawl API key? (yes/no)"
+
+If no:
+
+1. Direct them to: https://firecrawl.dev
+2. Explain:
+   - Sign up for free account
+   - Get 300 free credits to start
+   - Open-source project, can self-host if needed
+3. Guide to dashboard: "Go to your dashboard and copy your API key (starts with
+   'fc-...')"
+4. Wait for them to confirm they have the key
+
+If yes:
+
+1. Check if it's already in their environment: `echo $FIRECRAWL_API_KEY`
+2. If not set, ask them to provide it
+
+### 4. Configure Environment Variable
+
+Detect their shell:
+
+```bash
+echo $SHELL
+```
+
+Based on the shell, add the API key to the appropriate config file:
+
+**For zsh** (.zshrc):
+
+```bash
+echo 'export FIRECRAWL_API_KEY="fc-their-actual-key"' >> ~/.zshrc
+source ~/.zshrc
+```
+
+**For bash** (.bashrc or .bash_profile):
+
+```bash
+echo 'export FIRECRAWL_API_KEY="fc-their-actual-key"' >> ~/.bashrc
+source ~/.bashrc
+```
+
+Verify it's set:
+
+```bash
+echo $FIRECRAWL_API_KEY
+```
+
+### 5. Explain Available Scripts
+
+There are two scraping scripts in `.scripts/`:
+
+#### firecrawl-scrape.sh (Single URL)
+
+For scraping one URL at a time with control over the filename:
+
+```bash
+# Basic usage
+.scripts/firecrawl-scrape.sh "https://example.com/article" "00_Inbox/Clippings/article-name.md"
+
+# You choose the filename
+.scripts/firecrawl-scrape.sh "https://blog.com/post" "03_Resources/Articles/my-note.md"
+```
+
+#### firecrawl-batch.sh (Multiple URLs)
+
+For scraping multiple URLs at once with auto-generated filenames:
+
+```bash
+# Default: saves to 00_Inbox/Clippings/
+.scripts/firecrawl-batch.sh "https://url1.com" "https://url2.com" "https://url3.com"
+
+# Custom output directory
+.scripts/firecrawl-batch.sh -o "01_Projects/Research/" "https://url1.com" "https://url2.com"
+.scripts/firecrawl-batch.sh --output-dir "03_Resources/Articles/" "https://url1.com"
+```
+
+### 6. Create Necessary Folders
+
+Make sure the default Clippings folder exists:
+
+```bash
+mkdir -p 00_Inbox/Clippings
+```
+
+### 7. Test Connection
+
+Perform a test scrape to verify everything works:
+
+Ask: "Would you like to test with a sample article? (yes/no)"
+
+If yes:
+
+```bash
+.scripts/firecrawl-scrape.sh "https://example.com" "00_Inbox/test-scrape.md"
+```
+
+Check the output:
+
+- Look for: "✓ Scraped content saved to: 00_Inbox/test-scrape.md"
+- Verify the file has content: `wc -l 00_Inbox/test-scrape.md`
+- Show first few lines: `head -20 00_Inbox/test-scrape.md`
+
+If successful, offer to delete test file:
+
+```bash
+rm 00_Inbox/test-scrape.md
+```
+
+### 8. Show Research Workflows
+
+#### Workflow 1: Save Single Article
+
+```bash
+# User says: "Save this article to my vault: https://blog.example.com/great-post"
+# You run:
+.scripts/firecrawl-scrape.sh "https://blog.example.com/great-post" "00_Inbox/Clippings/great-post.md"
+```
+
+#### Workflow 2: Batch Research Collection
+
+```bash
+# User provides list of research URLs
+# You run:
+.scripts/firecrawl-batch.sh -o "01_Projects/AI-Research/" \
+  "https://paper1.com" \
+  "https://paper2.com" \
+  "https://paper3.com"
+```
+
+#### Workflow 3: Build Reference Library
+
+```bash
+# Save documentation and tutorials to Resources
+.scripts/firecrawl-batch.sh -o "03_Resources/Tech-Docs/" \
+  "https://docs.example.com/guide1" \
+  "https://docs.example.com/guide2"
+```
+
+#### Workflow 4: Reading List
+
+```bash
+# Quickly save interesting articles to read later
+.scripts/firecrawl-batch.sh \
+  "https://article1.com" \
+  "https://article2.com" \
+  "https://article3.com"
+# All saved to 00_Inbox/Clippings/ for later processing
+```
+
+### 9. Usage Tips
+
+**Best Practices:**
+
+1. **Organize as you save**: Use meaningful filenames and folders
+2. **Batch similar content**: Group related articles in same folder
+3. **Add frontmatter**: After scraping, add tags/metadata to notes
+4. **Link notes**: Create connections between related saved articles
+5. **Clean up**: Regularly process Clippings folder
+
+**Cost Management:**
+
+- Free tier: 300 credits
+- Each scrape uses ~1 credit
+- Monitor usage in Firecrawl dashboard
+- Can self-host for unlimited use
+
+**Common Patterns:**
+
+- Ask Claude: "Save this to my vault: [URL]"
+- Ask Claude: "Scrape these 10 articles into my Research folder: [URLs]"
+- Ask Claude: "Find and save the top 5 articles about [topic]"
+
+### 10. Success Message
+
+Show a summary:
+
+```
+✓ Firecrawl is now set up!
+
+Available scripts:
+  .scripts/firecrawl-scrape.sh <url> <output_file>
+  .scripts/firecrawl-batch.sh [-o dir] <url1> <url2> ...
+
+Usage with Claude:
+  "Save this article to my vault: https://example.com"
+  "Scrape these URLs to my Research folder: url1, url2, url3"
+
+Your research workflow is now supercharged!
+```
+
+## Troubleshooting
+
+Common issues:
+
+1. **"jq: command not found"**
+   - Install jq: `brew install jq` (macOS) or `apt-get install jq` (Linux)
+
+2. **"Failed to scrape content"**
+   - Check API key is set: `echo $FIRECRAWL_API_KEY`
+   - Verify key is valid at https://firecrawl.dev/dashboard
+   - Check credits remaining
+   - Some sites may block scraping
+
+3. **"Permission denied"**
+   - Make scripts executable: `chmod +x .scripts/firecrawl-*.sh`
+
+4. **Empty or incomplete files**
+   - Some sites have aggressive anti-scraping
+   - Try the URL directly in browser first
+   - Check if site requires authentication
+
+## Reference Documentation
+
+For detailed script documentation: `.scripts/README.md`
+
+For more web scraping options: https://docs.firecrawl.dev
+
+## Output
+
+Progress through each step, showing:
+
+- ✓ Checkmarks for completed steps
+- Clear examples of usage
+- Verification that scraping works
+- Workflow examples tailored to user's needs
+- Final success message with quick reference
+
+## Example Usage
+
+```
+/setup-firecrawl
+```
+
+Or with API key already configured:
+
+```
+/setup-firecrawl quick
+```
diff --git a/.claude/commands/setup-gemini.md b/.claude/commands/setup-gemini.md
new file mode 100644
index 00000000..d8f6ae38
--- /dev/null
+++ b/.claude/commands/setup-gemini.md
@@ -0,0 +1,211 @@
+---
+name: setup-gemini
+description:
+  Interactive setup wizard for Gemini Vision MCP server for image/video analysis
+allowed-tools: [Read, Write, Bash, AskUserQuestion]
+argument-hint:
+  "(optional) skip steps with 'quick' if you already have API key configured"
+---
+
+# Gemini Vision Setup Wizard
+
+This command guides you through setting up the Gemini Vision MCP server, which
+enables Claude to analyze images, videos, PDFs, and other visual content in your
+vault.
+
+## Task
+
+Set up Gemini Vision MCP server by:
+
+1. Checking prerequisites (Node.js v22+, pnpm, Claude Code)
+2. Getting a free Gemini API key from Google AI Studio
+3. Configuring the GEMINI_API_KEY environment variable
+4. Installing required dependencies
+5. Registering the MCP server with Claude
+6. Testing the connection
+
+## Process
+
+### 1. Welcome & Prerequisites Check
+
+Start by welcoming the user and explaining what Gemini Vision does:
+
+- Analyze images directly (screenshots, diagrams, photos)
+- Extract text from PDFs and scanned documents
+- Process videos (local files and YouTube URLs)
+- Auto-generate filenames based on content
+- Compare multiple images
+
+Then check prerequisites:
+
+```bash
+node --version  # Should be v22+
+pnpm --version  # Should be installed
+claude --version  # Should be installed
+```
+
+If any are missing, guide them to install:
+
+- Node.js: https://nodejs.org/ (v22+)
+- pnpm: `npm install -g pnpm`
+- Claude Code: https://claude.ai/code
+
+### 2. Get Gemini API Key
+
+Ask: "Do you already have a Gemini API key configured? (yes/no)"
+
+If no:
+
+1. Direct them to: https://aistudio.google.com/apikey
+2. Explain: "Click 'Create API Key' and copy the key (starts with 'AIzaSy...')"
+3. Wait for them to confirm they have the key
+
+If yes:
+
+1. Check if it's already in their environment: `echo $GEMINI_API_KEY`
+2. If not set, ask them to provide it
+
+### 3. Configure Environment Variable
+
+Detect their shell:
+
+```bash
+echo $SHELL
+```
+
+Based on the shell, add the API key to the appropriate config file:
+
+**For zsh** (.zshrc):
+
+```bash
+echo 'export GEMINI_API_KEY="their-actual-api-key"' >> ~/.zshrc
+source ~/.zshrc
+```
+
+**For bash** (.bashrc or .bash_profile):
+
+```bash
+echo 'export GEMINI_API_KEY="their-actual-api-key"' >> ~/.bashrc
+source ~/.bashrc
+```
+
+Verify it's set:
+
+```bash
+echo $GEMINI_API_KEY
+```
+
+### 4. Install Dependencies
+
+CRITICAL: Must install dependencies BEFORE registering MCP server!
+
+```bash
+pnpm install
+```
+
+This installs:
+
+- @google/generative-ai (Gemini API client)
+- @modelcontextprotocol/sdk (MCP server framework)
+- Other required dependencies
+
+If pnpm fails, fall back to npm:
+
+```bash
+npm install
+```
+
+### 5. Register MCP Server
+
+Ask: "Do you want project-scoped (just this vault) or user-scoped (all projects)
+installation?"
+
+**Project-scoped (recommended for this vault only)**:
+
+```bash
+claude mcp add --scope project gemini-vision node .claude/mcp-servers/gemini-vision.mjs
+```
+
+This will create .mcp.json. The MCP server will automatically inherit the
+GEMINI_API_KEY environment variable from your shell.
+
+**User-scoped (all projects)**:
+
+```bash
+claude mcp add --scope user gemini-vision node .claude/mcp-servers/gemini-vision.mjs
+```
+
+This will update ~/.claude.json. The MCP server will automatically inherit the
+GEMINI_API_KEY environment variable from your shell.
+
+**Important**: The API key is read from your shell environment (set in
+.zshrc/.bashrc), NOT stored in any config file. The MCP server inherits it
+automatically when Claude Code starts.
+
+### 6. Test Connection
+
+Important: User must open a NEW Claude Code window for MCP to connect!
+
+Guide them to:
+
+1. Exit current Claude session (or open new terminal)
+2. Start Claude again: `claude`
+3. Check connection: `/mcp`
+4. Look for: `gemini-vision ✔ connected`
+
+If connected, test with a sample command: "Try: 'Use gemini-vision to list
+available tools'"
+
+### 7. Success & Next Steps
+
+If everything works:
+
+- Congratulate them!
+- Show example commands:
+  - "Analyze this image: 05_Attachments/screenshot.png"
+  - "Extract text from this PDF: 05_Attachments/document.pdf"
+  - "Analyze this YouTube video: https://youtube.com/watch?v=..."
+  - "Suggest a filename for IMG_1234.jpg"
+
+If issues occur:
+
+- Direct them to troubleshooting guide:
+  `.claude/mcp-servers/GEMINI_VISION_QUICK_START.md`
+- Common issues:
+  - Dependencies not installed → run `pnpm install`
+  - Server not showing in /mcp → restart Claude Code
+  - API key not working → verify at https://aistudio.google.com/apikey
+
+## Troubleshooting Reference
+
+For detailed troubleshooting, refer to:
+`.claude/mcp-servers/GEMINI_VISION_QUICK_START.md`
+
+Common issues:
+
+1. "Cannot find package '@modelcontextprotocol/sdk'" → Run `pnpm install`
+2. Server shows "failed" → Check API key is set: `echo $GEMINI_API_KEY`
+3. Tools don't work → Verify API key is valid, restart terminal to reload env
+   vars
+4. Server not listed in /mcp → Must restart Claude Code after adding server
+
+## Output
+
+Progress through each step, showing:
+
+- ✓ Checkmarks for completed steps
+- Clear instructions for manual steps
+- Verification of each component
+- Final success message with usage examples
+
+## Example Usage
+
+```
+/setup-gemini
+```
+
+Or with API key already configured:
+
+```
+/setup-gemini quick
+```