# TwoSplit

> Use this tool when you need to generate more comprehensive and nuanced AI responses by comparing multiple perspectives. TwoSplit sends a prompt to two Claude models, then uses a third instance to select or combine the best elements, delivering an optimized answer along with original outputs and source attribution. It's ideal for users seeking enhanced accuracy and depth in AI-generated responses, supporting various Claude models like Opus, Sonnet, and Haiku.

Canonical page: https://skillsregistry.net/skills/lazer-twosplit  
JSON: https://api.skillsregistry.net/v1/skills/lazer-twosplit

## Description

Twosplit MCP Server leverages multiple Claude instances to provide enhanced AI responses. It sends the same prompt to two separate Claude models and uses a third instance to either select the best response or combine elements from both, delivering a final optimized answer. The server supports various Claude models including Opus, Sonnet, and Haiku, and returns not only the final response but also the original outputs from both AIs along with source attribution. This implementation is particularly valuable for users seeking more comprehensive or nuanced answers by comparing multiple AI perspectives within a single query.

## Trust

- **Trust score (0–1):** 0.65
- **Verification tier:** scanned
- **Last scanned:** 2026-09-02

## Facts

- **Version:** 1.0.0
- **Skill type:** atomic
- **Execution layer:** mcp-remote
- **Runtime environment:** api
- **Category:** ai-ml
- **Updated:** 2026-09-02

## Source

- **Source listing:** [PulseMCP](https://www.pulsemcp.com/servers/lazer-twosplit)
- **Repository:** <https://github.com/lazerthings/twosplit>

## Use it

Resolve this record through the SkillsRegistry MCP server (no auth, read-only):

```
claude mcp add --transport http --scope user skillsregistry https://api.skillsregistry.net/mcp
```

```json
{
  "jsonrpc": "2.0",
  "id": 1,
  "method": "tools/call",
  "params": {
    "name": "get_skill",
    "arguments": {
      "slug": "lazer-twosplit"
    }
  }
}
```

REST: `GET https://api.skillsregistry.net/v1/skills/lazer-twosplit` · pull for local use: `GET https://api.skillsregistry.net/v1/skills/lazer-twosplit/pull`

---
SkillsRegistry indexes agent skills from public registries and GitHub. Skills we have analysed are scanned with Circle-IR and scored on six dimensions; each listing states its scan coverage. More: https://skillsregistry.net/llms.txt