API Reference

Complete reference documentation for the Darbot Deepmind MCP Server tool and its parameters.

Tool: darbot_deepmind

The primary tool for sophisticated, step-by-step reasoning through Darbot's deepmind thinking framework.

Description

This tool facilitates complex problem-solving and analysis through a structured thinking process. It supports dynamic thought revision, multi-path reasoning with branching logic, and adaptive planning that adjusts based on problem complexity.

Parameters

Required Parameters

Optional Parameters

Response Format

The tool returns a structured JSON response with the following fields:

{
  "thoughtNumber": 5,
  "totalThoughts": 10,
  "nextThoughtNeeded": true,
  "branches": ["serverless-approach", "microservices-approach"],
  "thoughtHistoryLength": 5,
  "isRevision": false,
  "branchId": "serverless-approach",
  "branchFromThought": 3
}

Usage Examples

Example 1: Basic Sequential Reasoning

// First thought
{
  "thought": "Analyzing requirements for e-commerce platform",
  "nextThoughtNeeded": true,
  "thoughtNumber": 1,
  "totalThoughts": 8
}

// Second thought
{
  "thought": "Identifying key scalability challenges",
  "nextThoughtNeeded": true,
  "thoughtNumber": 2,
  "totalThoughts": 8
}

Example 2: Revising Previous Thinking

{
  "thought": "Revising database choice - NoSQL better for this use case",
  "nextThoughtNeeded": true,
  "thoughtNumber": 6,
  "totalThoughts": 10,
  "isRevision": true,
  "revisesThought": 3
}

Example 3: Branching to Explore Alternatives

{
  "thought": "Exploring serverless architecture as alternative",
  "nextThoughtNeeded": true,
  "thoughtNumber": 7,
  "totalThoughts": 12,
  "branchFromThought": 4,
  "branchId": "serverless-alternative"
}

Example 4: Adjusting Scope

{
  "thought": "Realizing this requires more detailed security analysis",
  "nextThoughtNeeded": true,
  "thoughtNumber": 8,
  "totalThoughts": 15,
  "needsMoreThoughts": true
}

Example 5: Final Thought

{
  "thought": "Conclusion: Microservices with event-driven architecture",
  "nextThoughtNeeded": false,
  "thoughtNumber": 12,
  "totalThoughts": 12
}

Validation Rules

Schema Validation

• thought must be a non-empty string
• thoughtNumber must be a positive integer
• totalThoughts must be a positive integer
• revisesThought (if present) must be a positive integer
• branchFromThought (if present) must be a positive integer

Business Logic Validation

• If isRevision is true, revisesThought must be provided
• revisesThought must be less than thoughtNumber
• If branchId is provided, branchFromThought should be provided
• thoughtNumber cannot be greater than totalThoughts (server auto-adjusts if needed)

Error Responses

When validation fails, the tool returns an error response:

{
  "error": "Validation error: revisesThought must be less than thoughtNumber",
  "status": "failed"
}

Reasoning Guidelines

When to Use Revisions

• New information invalidates previous assumptions
• Discovered a flaw in earlier reasoning
• Need to reconsider a decision with additional context
• Realized a better approach after further analysis

When to Use Branches

• Exploring fundamentally different solution approaches
• Comparing trade-offs between alternatives
• Investigating multiple hypotheses in parallel
• Examining what-if scenarios

When to Adjust Total Thoughts

• Problem is more complex than initially estimated
• Found solution faster than expected
• Scope of analysis has changed
• New requirements emerged during reasoning

Thought Content Best Practices

Good Thought Examples

✓ "Analyzing database options: PostgreSQL vs MongoDB"
✓ "Security concern: Need to address authentication flow"
✓ "Trade-off: Consistency vs availability in distributed system"
✓ "Hypothesis: Caching layer will reduce latency by 50%"

Avoid

✗ "Thinking..."
✗ "Next step"
✗ "Continuing analysis"
✗ "" (empty thoughts)

Thought Content Guidelines

• Be specific about what you're analyzing or deciding
• Include key insights, not just process statements
• Reference concrete details when relevant
• State hypotheses clearly when generating them
• Note important trade-offs or constraints

Integration Examples

JavaScript/TypeScript

const response = await callTool("darbot_deepmind", {
  thought: "Analyzing system architecture requirements",
  nextThoughtNeeded: true,
  thoughtNumber: 1,
  totalThoughts: 10
});

console.log(response.thoughtHistoryLength); // 1
console.log(response.branches); // []

Python (Pseudo-code)

response = mcp_client.call_tool(
    "darbot_deepmind",
    {
        "thought": "Analyzing system architecture requirements",
        "nextThoughtNeeded": True,
        "thoughtNumber": 1,
        "totalThoughts": 10
    }
)

print(response["thoughtHistoryLength"])  # 1
print(response["branches"])  # []

Performance Notes

Memory Usage

• Each thought stored in memory (~1KB average)
• 100 thoughts ≈ 100KB memory
• Branch tracking adds minimal overhead
• Consider session breaks for very long chains (1000+ thoughts)

Response Time

• Typical response time: <5ms
• Validation: O(1) complexity
• Logging adds ~2-3ms when enabled
• Network latency dominates total time

Troubleshooting

Common Errors

Validation Error: thought must be non-empty

Cause: Empty or missing thought parameter

Solution: Provide a meaningful description of the current thinking step

Validation Error: revisesThought must be less than thoughtNumber

Cause: Attempting to revise a future or current thought

Solution: Only revise past thoughts (revisesThought < thoughtNumber)

Validation Error: revisesThought required when isRevision is true

Cause: Set isRevision to true without specifying which thought to revise

Solution: Include the revisesThought parameter

Next Steps