Resource Center

Overview

AgentPsy is a tool designed for multi-dimensional personality assessment (Big Five), supporting a wide range of evaluators from OpenAI, Google, and Anthropic to local Ollama instances. It is configuration-driven, generates detailed analysis reports, and features a powerful batch analysis capability for large datasets.

Source Code: https://github.com/ptreezh/AgentPsyAssessment.git

License: This project is licensed under the Apache License 2.0. See LICENSE for more information.

Batch Analysis CLI

The batch analysis script offers flexible control:

# 1. Get file statistics
python ultimate_batch_analysis.py --stats

# 2. Run a quick test (5 files)
python ultimate_batch_analysis.py --quick

# 3. Analyze a specific model series
python ultimate_batch_analysis.py --filter deepseek

# 4. Analyze 10 files from a series with a specific evaluator
python ultimate_batch_analysis.py --filter deepseek --sample 10 --evaluators ollama_llama3

Output Structure

Results are saved in batch_analysis_results/, containing summaries and detailed reports for each batch.

Step 1: Clone the Repository

git clone https://github.com/ptreezh/AgentPsyAssessment.git
cd AgentPsy

Step 2: Install Dependencies

Core dependencies are listed in requirements.txt. Optional providers require separate installation.

# Install core dependencies
pip install -r requirements.txt

# Optional: Install other providers
pip install google-generativeai anthropic dashscope

Step 3: Configure Environment

Copy .env.example to .env and fill in your API keys. For local analysis, ensure Ollama is running.

# .env file content
PROVIDER=local # or 'cloud'
LOCAL_API_BASE=http://localhost:11434/v1
LOCAL_MODEL_ID=llama3

# Cloud Keys (if needed)
GEMINI_API_KEY=your_gemini_key
...

Step 4: Run an Analysis

Use the batch analysis script to start.

python ultimate_batch_analysis.py --quick