# Outline AI Workflow

**Date:** 2026-02-01 21:08
**Context:** Setup MCP server for AI interaction with Outline wiki

---

## Quick Reference

| Item | Value |
|------|-------|
| **Outline URL** | http://192.168.1.3:3080 |
| **API URL** | http://192.168.1.3:3080/api |
| **MCP Server** | `mcp-outline` (via uvx) |
| **NUC Docs Collection** | `2a42945b-1a4f-4c92-add5-dfa147ef3f56` |
| **API Key** | `ol_api_nIjWn2lJn4Ho42kKcivcqTHjyy8Vg9ycxmchHa` |
| **Key Name** | Claude MCP v2 (full access) |
| **Key Expires** | Mar 03, 2026 |

---

## MCP Tools Available

After restarting Claude Code (`/mcp` to verify):

```
mcp__outline__search_documents       # Search by keywords
mcp__outline__list_collections       # List all collections
mcp__outline__read_document          # Get document content
mcp__outline__create_document        # Create new document
mcp__outline__update_document        # Update document
mcp__outline__move_document          # Move to different collection
mcp__outline__archive_document       # Archive document
mcp__outline__delete_document        # Delete document
mcp__outline__export_document        # Export as markdown
mcp__outline__batch_create_documents # Create multiple docs
mcp__outline__ask_ai_about_documents # AI queries on docs
```

---

## Common Workflows

### 1. Save Important Finding to Wiki

```python
mcp__outline__create_document(
    title="<Descriptive Title>",
    text="<Markdown content>",
    collectionId="2a42945b-1a4f-4c92-add5-dfa147ef3f56",
    publish=True
)
```

### 2. Search Before Solving

```python
# Check if solution already documented
results = mcp__outline__search_documents(query="tailscale funnel")
```

### 3. Update Existing Doc

```python
mcp__outline__update_document(
    id="<doc-id>",
    text="<new content>",
    append=False  # True to append instead of replace
)
```

### 4. Find Doc by Title

```python
doc_id = mcp__outline__get_document_id_from_title(title="Architecture")
```

---

## REST API Fallback

When MCP unavailable:

```bash
# Create document
curl -X POST 'http://192.168.1.3:3080/api/documents.create' \
  -H 'Authorization: Bearer ol_api_cs243A8BKmEW6yAdrMp05PxTILMBjPofIQcVcM' \
  -H 'Content-Type: application/json' \
  -d '{
    "title": "Title",
    "text": "Content",
    "collectionId": "2a42945b-1a4f-4c92-add5-dfa147ef3f56",
    "publish": true
  }'

# Search
curl -X POST 'http://192.168.1.3:3080/api/documents.search' \
  -H 'Authorization: Bearer ol_api_cs243A8BKmEW6yAdrMp05PxTILMBjPofIQcVcM' \
  -H 'Content-Type: application/json' \
  -d '{"query": "search term"}'
```

---

## Config Location

`~/.claude/settings.json`:

```json
{
  "mcpServers": {
    "outline": {
      "command": "uvx",
      "args": ["mcp-outline"],
      "env": {
        "OUTLINE_API_KEY": "ol_api_cs243A8BKmEW6yAdrMp05PxTILMBjPofIQcVcM",
        "OUTLINE_API_URL": "http://192.168.1.3:3080/api"
      }
    }
  }
}
```

---

## When to Use

**Save to Outline (persistent docs):**
- Infrastructure guides
- Configuration docs
- Troubleshooting procedures
- Architecture decisions

**Save to .artifacts/ (session notes):**
- API keys/tokens generated
- Temporary findings
- Session-specific configs

---

## Generate New API Key

1. Go to http://192.168.1.3:3080/settings/api-and-apps
2. Click "New API key..."
3. Set scopes: `documents.create documents.update documents.read collections.read`
4. Copy immediately (shown once)
5. Update `~/.claude/settings.json`

---

## Related

- MCP Server: https://github.com/Vortiago/mcp-outline
- Outline API: https://www.getoutline.com/developers
- Full guide: `docs/outline-ai-workflow.md`