Quick Start Guide¶
Goal
Get up and running with AI project management in 15 minutes
This guide walks you through setting up the system and creating your first project using AI commands while observing the results visually in Obsidian.
Prerequisites¶
- Obsidian installed
- Node.js 18+ installed
- Claude Desktop or Cursor IDE
- An empty Obsidian vault (or create a new one)
Step 1: Install MCP Server (5 minutes)¶
- Open Claude Desktop configuration file:
- macOS:
~/Library/Application Support/Claude/claude_desktop_config.json -
Windows:
%APPDATA%\Claude\claude_desktop_config.json -
Add the MCP server configuration:
{
"mcpServers": {
"obsidian": {
"command": "npx",
"args": ["-y", "obsidian-accomplishments-mcp"],
"env": {
"VAULT_PATH": "/absolute/path/to/your/vault",
"DEFAULT_CANVAS": "main.canvas"
}
}
}
}
-
Replace
/absolute/path/to/your/vaultwith your actual vault path -
Restart Claude Desktop
- Open Cursor Settings → MCP Servers
- Add new server with the same configuration as above
- Restart Cursor
Verify
Ask Claude/Cursor: "List available MCP tools" - you should see obsidian tools listed
Step 2: Install Plugin (5 minutes)¶
Note
Plugin is under review. Install from source for now.
- Clone and build the plugin:
git clone https://github.com/ostanlabs/obsidian_plugin.git
cd obsidian_plugin
npm install
npm run build
- Copy to your vault:
mkdir -p /path/to/vault/.obsidian/plugins/canvas-project-manager
cp main.js manifest.json styles.css /path/to/vault/.obsidian/plugins/canvas-project-manager/
- In Obsidian:
-
Settings → Community Plugins → Enable Canvas Project Manager
-
Create a canvas file:
- Create new file:
main.canvas
Verify
Open command palette (Ctrl/Cmd+P) and search for "Project Canvas" - you should see plugin commands
Step 3: Create Your First Project with AI (5 minutes)¶
Now let's create a simple learning project using AI and watch it appear on the canvas!
3.1 Create the Learning Milestone¶
Ask your AI assistant:
Create a milestone called "Learn Obsidian Project Management" with target date
one week from today, workstream "learning", and priority "High"
Expected Response:
✓ Created M-001: Learn Obsidian Project Management
- Target: [date]
- Workstream: learning
- Priority: High
3.2 Add Stories¶
Ask your AI:
Create two stories under M-001:
1. "Set up the system" - status "Completed"
2. "Create first real project" - status "Not Started"
Both in workstream "learning"
Expected Response:
✓ Created S-001: Set up the system (parent: M-001)
✓ Created S-002: Create first real project (parent: M-001)
3.3 Add Tasks¶
Ask your AI:
Create three tasks under S-002:
1. "Define project goals" - status "Open"
2. "Break down into milestones" - status "Open"
3. "Create initial stories" - status "Open"
All in workstream "learning"
Expected Response:
✓ Created T-001: Define project goals (parent: S-002)
✓ Created T-002: Break down into milestones (parent: S-002)
✓ Created T-003: Create initial stories (parent: S-002)
Step 4: Visualize on Canvas (2 minutes)¶
Now let's see what we created!
-
Open
main.canvasin Obsidian -
Run command (Ctrl/Cmd+P):
-
"Project Canvas: Populate from vault"
-
Run command again:
- "Project Canvas: Reposition nodes"
You should see:
graph TD
M[M-001: Learn Obsidian PM<br/>Target: date | Priority: High]
S1[S-001: Set up<br/>✓ Complete]
S2[S-002: Create<br/>Not Started]
T1[T-001: Define goals]
T2[T-002: Break down]
T3[T-003: Create stories]
M --> S1
M --> S2
S2 --> T1
S2 --> T2
S2 --> T3
style M fill:#e3f2fd
style S1 fill:#c8e6c9
style S2 fill:#fff9c4
style T1 fill:#ffe0b2
style T2 fill:#ffe0b2
style T3 fill:#ffe0b2
Step 5: Iterate and Observe (3 minutes)¶
Now let's practice the AI → Visual workflow:
Check Status¶
Ask AI:
Take Action¶
Ask AI:
Observe Change¶
- In Obsidian, refresh canvas (close and reopen, or run "Populate from vault" again)
- See T-001 now shows as complete with different styling
Report Progress¶
Ask AI:
Move Forward¶
Ask AI:
Refresh canvas → See T-002 status updated
🎉 Success!¶
You've now:
- Set up MCP Server and Plugin
- Created a 3-tier project structure via AI
- Visualized it on Obsidian canvas
- Iterated: AI action → Visual review
Next Steps¶
-
User Guide
Learn all features and workflows
-
First Project
Create your first real project
-
Sample Project
Explore a complete example project
Troubleshooting¶
No entities found
- Check that
VAULT_PATHin MCP config points to the correct vault - Verify files were created in vault (check
milestones/,stories/,tasks/folders)
Canvas is empty
- Make sure you ran "Populate from vault" command
- Check that
main.canvasfile exists - Look for errors in Developer Console (Ctrl/Cmd+Shift+I)
MCP tools not available
- Restart Claude Desktop / Cursor
- Check MCP config file syntax (valid JSON)
- Verify Node.js is installed:
node --version
For more help, see FAQ