Skip to content

CRISP-T Web UI - Quick Start Guide

Installation

Step 1: Install CRISP-T with Copilot support

pip install crisp-t[copilot]

This installs: - CRISP-T core package - Quart web framework - GitHub Copilot SDK - All necessary dependencies

Step 2: Install and configure GitHub Copilot CLI

Follow the official guide: https://docs.github.com/en/copilot/how-tos/set-up/install-copilot-cli

Quick setup:

# Install GitHub CLI if not already installed
# See: https://cli.github.com/

# Login to GitHub
gh auth login

# Verify Copilot CLI
copilot --version

Quick Start

Launch the Web UI

# Start on default port (5000)
crisp-ui

# Or specify a custom port
crisp-ui --port 8080

Open your browser to: http://127.0.0.1:5000

First Session

  1. Configure Settings (left panel):
  2. Select AI model (e.g., GPT-5)
  3. Set data path (e.g., ./data)

  4. Optionally configure advanced settings

  5. Start Session:

  6. Click "Start Session" button

  7. Wait for green status indicator

  8. Start Chatting:

  9. Type your request in the chat box
  10. Press Enter or click Send
  11. Watch the AI respond in real-time

Example Workflow

Basic Analysis

You: Import data from ./interviews folder

AI: I'll import your data using CRISP-T...
[Runs: crisp --source ./interviews --out corpus]

You: What's in the corpus?

AI: Let me check...
[Runs: crispt --inp corpus --print]

You: Run topic modeling with 5 topics

AI: I'll perform topic modeling...
[Runs: crisp --inp corpus --topics --num 5 --assign --out corpus]

You: Create a word cloud

AI: Creating visualization...
[Runs: crispviz --inp corpus --wordcloud --out viz]

Advanced Analysis

You: Import CSV data from ./survey with "feedback" as the text column,
     run sentiment analysis, and link it to the satisfaction_score variable

AI: I'll perform a comprehensive mixed-methods analysis...
[Runs multiple commands in sequence]

Using Local Models (Ollama)

Setup

  1. Install Ollama: https://ollama.ai
  2. Pull a model: bash ollama pull deepseek-coder-v2:16b

Configure in UI

  1. Check "Use Custom Provider"
  2. Set:
  3. Provider Type: OpenAI
  4. Base URL: http://localhost:11434/v1
  5. API Key: (leave empty)
  6. Model: Enter deepseek-coder-v2:16b
  7. Start Session

Troubleshooting

"Copilot SDK not available"

Solution: Install with copilot support

pip install crisp-t[copilot]

"Failed to create session"

Cause: Copilot CLI not installed or not logged in

Solution:

# Check if installed
copilot --version

# If not, install GitHub CLI and Copilot CLI
# Then login
gh auth login

Port Already in Use

Solution: Use different port

crisp-ui --port 8080

Commands Not Found

Cause: CRISP-T CLI tools not in PATH

Solution: Reinstall CRISP-T

pip install crisp-t[ml]

Tips

  1. Start Simple: Begin with basic commands to familiarize yourself
  2. Be Specific: Provide clear instructions to the AI
  3. Use Examples: The welcome screen shows example prompts
  4. Check Results: Review command outputs in the chat
  5. Save Work: Use --out flags to save intermediate results

Next Steps

  • Read the full documentation: docs/ui.md
  • Explore CRISP-T CLI: docs/cheatsheet.md
  • Try example workflows: docs/DEMO.md
  • Learn about the crisp-cli skill: .agents/skills/crisp-cli/SKILL.md

Getting Help

  • GitHub Issues: https://github.com/dermatologist/crisp-t/issues
  • Documentation: https://dermatologist.github.io/crisp-t/
  • CRISP-T Wiki: https://github.com/dermatologist/crisp-t/wiki

Happy analyzing! 🔍