MCP Server Documentation

Search the knowledge base for relevant information. Returns the most semantically similar documents to your query, ranked by relevance score.

Parameters

Example Usage

// In Claude, Gemini, or any MCP client:
"Search my knowledge base for authentication best practices"

// The AI will call:
search_knowledge_base({
  "query": "authentication best practices",
  "limit": 5
})

// Returns:
{
  "results": [
    {
      "content": "Always use bcrypt with cost factor 12+...",
      "source": "security-docs",
      "score": 0.92,
      "metadata": { "section": "passwords" }
    },
    // ... more results
  ]
}

Add new content to the knowledge base. The content is automatically chunked into semantic segments, embedded using OpenAI's text-embedding-3-small model, and indexed for retrieval.

Parameters

Example Usage

// In Claude, Gemini, or any MCP client:
"Add this to my knowledge base: Our API uses JWT tokens 
with a 15-minute expiration for access tokens."

// The AI will call:
add_memory({
  "content": "Our API uses JWT tokens with a 15-minute expiration for access tokens.",
  "source": "api-documentation",
  "metadata": {
    "category": "authentication",
    "updated": "2026-02-02"
  }
})

// Returns:
{
  "success": true,
  "chunks": 1,
  "message": "Content stored successfully"
}

CClaude CLI - Querying the Knowledge Base

$ claude

You: What do we know about rate limiting in our API?

Claude: I'll search your knowledge base for information about rate limiting.

[Calling tool: search_knowledge_base]
[Arguments: {"query": "rate limiting API", "limit": 5}]

Based on your knowledge base, here's what I found about rate limiting:

1. **Rate Limits by Plan**
   - Starter: 100 requests/minute
   - Pro: 1,000 requests/minute
   - Enterprise: Custom limits

2. **Rate Limit Headers**
   Your API returns these headers:
   - X-RateLimit-Limit: Maximum requests allowed
   - X-RateLimit-Remaining: Requests left in window
   - X-RateLimit-Reset: Unix timestamp when limit resets

3. **Handling 429 Errors**
   When rate limited, implement exponential backoff...

Would you like more details on any of these topics?

GGemini CLI - Adding to Memory

$ gemini

You: I just finished implementing the new webhook system. Add this 
documentation to my knowledge base:

Webhook Events:
- user.created: Fired when a new user signs up
- user.updated: Fired when user profile changes  
- payment.completed: Fired after successful payment
- subscription.changed: Fired when plan changes

All webhooks include a signature header for verification.
Retry policy: 3 attempts with exponential backoff.

Gemini: I'll store this webhook documentation in your knowledge base.

[Calling tool: add_memory]
[Arguments: {
  "content": "Webhook Events:\n- user.created: Fired when...",
  "source": "webhook-documentation",
  "metadata": {
    "type": "api-reference",
    "category": "webhooks",
    "author": "engineering"
  }
}]

Done! I've added the webhook documentation to your OrbitalMCP knowledge 
base. It's now indexed and searchable. The content was split into 2 
semantic chunks for optimal retrieval.

Would you like me to also add example payloads for each event type?

Connection refused or timeout

Ensure your API key is valid and has the correct permissions. Check that the Authorization header is properly formatted as 'Bearer YOUR_API_KEY'.

Tools not appearing in my AI client

Restart your MCP client after adding the configuration. Some clients require a full restart to detect new MCP servers.

Search returns no results

Make sure you've ingested content first. Use the add_memory tool or REST API to populate your knowledge base before searching.

Rate limiting errors

The MCP server respects the same rate limits as the REST API. If you're hitting limits, consider upgrading your plan or implementing request batching.