VersperClaw: From Idea to Search, Experiment and Paper - Not Only Personal AI Assistant

Quick Start

git clone https://github.com/math-inc/VersperClaw.git
cd VersperClaw

# make sure you have installed uv and creat .venv via uv venv
uv pip install -r requirements.txt # or uv sync
source .venv/bin/activate
uv pip install -e .

# use codex, openrouter, or claude remote model via api
versper setup

# if use vllm or llama.cpp in local model for inference 
# for example use llama.cpp
llama-server \
  --model unsloth/Qwen3.5-4B-GGUF/Qwen3.5-4B-UD-Q4_K_XL.gguf \
  --mmproj unsloth/Qwen3.5-4B-GGUF/mmproj-F16.gguf \
  --seed 3407 \
  --temp 1.0 \
  --top-p 0.95 \
  --min-p 0.01 \
  --top-k 40 \
  -c 49152 \
  --port 8001 \
  --chat-template-kwargs '{"enable_thinking":true}' \
  --host 0.0.0.0

# check model name and pass inference
curl http://localhost:8001/v1/models

curl http://localhost:8001/v1/chat/completions \
  -H "Content-Type: application/json" \
  -d '{
    "model": "qwen",
    "messages": [
      {"role": "user", "content": "hello"}
    ]
  }'

cd ~/.versper && nvim config.yaml
# model:
#  api_key: sk-no-key-required
#  base_url: http://localhost:8001/v1
#  default: unsloth/Qwen3.5-4B-GGUF
#  provider: custom

versper doctor # make sure all checks passed

Add-on

if ban mcp in ~/.versper/.env:
export VERSPER_BROWSER_USE_MCP=0

# live chrome control
npm install -g chrome-devtools-mcp@latest
npm install -g agent-browser@latest

# make sure you have installed chrome
google-chrome-stable

# fill it in the search bar and click the box
chrome://inspect/#remote-debugging

# check chrome connect
versper
/browser status

Examples

1.CLI Scientific Research

2.Web-Search & Deep-Research

3.Live Chrome Control

4.Gateway Chat

# telegram, wechat and whatsapp
versper setup gateway

# then start listen
versper gateway

5.Research Workflow

# in arch linux and must installed uv for experiment

1️⃣ Skill（工具层） 2️⃣ Agent（决策层） 3️⃣ Workflow / Pipeline（流程层）

skills/ search summarize backtest

workflows/ research/ trading/

agents/ planner_agent execution_agent