# Capacities MCP for AI Agents MCP

> Capacities MCP connects your AI agent directly to your personal knowledge graph, letting you build structured data and manage complex information within Capacities. You can capture web links with automatic previews, create new typed objects based on specific rules, append quick notes, and search across all your spaces using only natural conversation.

## Overview
- **Category:** productivity
- **Price:** Free
- **Tags:** knowledge-graph, object-oriented, personal-knowledge-management, unstructured-data, ai-agents, data-modeling

## Description

Think of this MCP as giving your AI agent a full set of hands inside your personal knowledge workspace. It lets you move beyond simple text files; you build interconnected, object-based graphs. Instead of copy-pasting links into a document and hoping they'll be found later, you can tell the agent to save a web URL, and it automatically populates all the metadata and dynamic previews for you. You don't have to manually structure every thought. Need to capture notes from a meeting? Just ask your agent to append markdown text directly into today’s daily log. If you need to track system architecture or product components, the agent can create entirely new typed objects that adhere strictly to your predefined schema. It takes natural language and turns it into structured data points. To start, just connect this MCP through Vinkius; then, let your preferred AI client handle the heavy lifting, letting you build deep knowledge blocks dynamically without ever leaving your chat interface.

## Tools

### list_spaces
Retrieves a list of every top-level container, or 'space,' in your Capacities account.

### get_space_info
Pulls detailed configuration for a specific space, including its defined object types and structure rules.

### get_structures
Lists all available object type definitions (structures) within a selected Capacities space.

### lookup
Searches for specific content nodes or objects across your graph by title or keyword.

### save_weblink
Saves a given web URL into Capacities as a Weblink object, generating automatic previews and metadata.

### save_to_daily_note
Appends plain text or Markdown payload directly to the daily note log, linking the content block permanently.

### create_object
Builds a new typed object within a space by ensuring it complies with predefined graph rules and structure parameters.

### get_object
Retrieves the full data payload for a specific, existing object using its unique ID.

### save_media
Locates and attaches an image or media file, binding it directly to another specific record within your graph.

### add_tag
Adds a structural tag to any item, helping you dynamically group related records and knowledge items together.

## Prompt Examples

**Prompt:** 
```
Find all notes related to 'Q3 Marketing Strategy' in my 'Work' space and summarize them into a bullet list.
```

**Response:** 
```
**Search Results for Q3 Marketing Strategy:**

*   **Object: Campaign Launch Plan (ID: O90s...)**
    *   Status: Drafted
    *   Key Dates: Sept 1 - Oct 31
*   **Object: Target Audience Profile (ID: O72t...)**
    *   Primary Segment: Small Businesses
    *   Pain Point: Lack of digital visibility.

Overall, the consensus is to focus on digital ads and update the profile object with more specific pain points. Should I attach this search result summary as a new note?
```

**Prompt:** 
```
Take this article link about quantum computing and save it to my 'Research' space.
```

**Response:** 
```
**Weblink Saved Successfully!**

The resource has been added to the Research space (ID: X74n...). Here are the automatic details:

*   **Title:** Quantum Computing Breakthroughs in 2025
*   **Source:** Nature Magazine
*   **Preview Snapshot:** [Image of quantum circuit]
*   **Metadata Tagged:** #Physics, #FutureTech

The dynamic preview is live and ready for you to review.
```

**Prompt:** 
```
I wrote a block of code fixing the login bug. Add it as markdown to my daily note.
```

**Response:** 
```
Done. I've appended your Python fix and a quick summary into today’s log in the main workspace. It looks like this:

```python
def fix_login(user):
    # Bug fixed: Added session validation check
    return True
```

*Note: This entry is permanently mapped to your daily activity and securely stored with a 'Bugfix' tag.*
```

## Capabilities

### Map and view all personal spaces
List every top-level container in your Capacities account so you know exactly where information lives.

### Get detailed space structures
Retrieve the full metadata on a specific space, including all available object types and their property definitions.

### Search for content in your graph
Execute rapid keyword searches across a specified space to locate active nodes or objects by title.

### Save web links with previews
Inject external URLs into your knowledge base as Weblink objects, which automatically generate dynamic previews and metadata.

### Append quick markdown notes
Send short thoughts or summaries using Markdown text directly to the daily note log, linking content blocks permanently.

### Create new custom objects
Generate a typed object within a space that strictly adheres to your established graph rules and schema parameters.

### Organize with tags and media
Attach images or add structural tags to existing records, instantly grouping related items in your knowledge graph.

## Use Cases

### Building a Project Scope Document
A developer needs to document a new microservice. They prompt the agent to create a typed object using `create_object`, defining its parameters (owner, endpoints, dependencies). The agent enforces the correct structure, building a reliable record immediately.

### Consolidating Meeting Learnings
A content strategist attended several meetings. Instead of making five separate notes, they ask their agent to process all raw links and meeting minutes, using `save_weblink` and `add_tag` to create a single, linked 'Client Brief' object.

### Researching Interconnected Ideas
A student needs to build a literature review. They use the agent to execute `lookup` on their 'Reading Material' space using keywords from several sources, gathering all relevant notes and links into one summary.

### Daily Thought Capture
At the end of a busy day, an executive can simply ask the agent to append today’s key takeaways using `save_to_daily_note`—a quick markdown bullet list that links them back to specific project objects.

## Benefits

- Stop losing context. When you use the `save_weblink` tool, your agent doesn't just paste a URL; it generates dynamic previews and metadata right into your space.
- Keep everything chronologically organized by appending notes. The `save_to_daily_note` tool lets you log quick thoughts or meeting summaries directly to today’s entry without leaving the chat.
- Move beyond simple documents. Use `create_object` to build new, typed entities that strictly follow your system's rules, making data reliable and queryable.
- Find what you need instantly. The agent uses `lookup` across specific spaces to track down active nodes or properties based on keywords, saving hours of manual searching.
- Keep your graph clean. With the ability to `add_tag`, you can quickly categorize related items after they've been saved via any process, keeping your knowledge organized by relation.

## How It Works

The bottom line is, you feed the AI agent a natural language prompt, and it translates that into precise actions within your object-based system.

1. Subscribe to this MCP on Vinkius.
2. Navigate to your Capacities app Settings and grab your Personal Token.
3. Start giving commands through Claude or Cursor, asking it to fetch specific data points or create structured objects in your knowledge base.

## Frequently Asked Questions

**How does the Capacities MCP help me connect my web research?**
It lets you save external links as structured Weblink objects. Instead of just pasting a URL, it captures the page's full metadata and generates dynamic previews right in your knowledge graph for easy review.

**Can I use Capacities MCP to track my project components?**
Yes. You can instruct the agent to build new typed objects that strictly follow your defined schemas, ensuring every component record is consistent and queryable across your space.

**What if I need to find old notes in my graph?**
The MCP allows you to run targeted lookups. You can ask the agent to search for content by keyword or object type within a specific space, quickly retrieving nodes that might otherwise get lost.

**Is Capacities MCP only for big teams?**
No. It's built for individual knowledge workers and researchers. You can use it to manage personal thought flows and build your own comprehensive, object-based private knowledge graph.

**Does the MCP help me keep a daily log of my thoughts?**
Absolutely. You can tell the agent to append text or code directly into today's daily note using markdown syntax. This keeps your immediate thoughts logged and permanently linked within your knowledge base.