00 · Installation
Cognithor runs on your own machine — no cloud account, no API key required.
Prerequisites
- Python 3.12 or newer (
python --version) - Ollama ≥ 0.4.0 (local LLM runtime) — ollama.com
- ~2 GB of free disk space (for the
qwen3:8bmodel) - Internet connection for the one-time
pip install+ model download
Time: 3 minutes
End state: cognithor --version prints 0.93.0, and ollama list shows at least qwen3:8b.
Option A — Windows One-Click Installer (recommended on Windows)
- Download
CognithorSetup-0.93.0-x64.exefrom github.com/Alex8791-cyber/cognithor/releases. - Double-click, "Next", "Install". The installer bundles a portable Python runtime + ffmpeg + an Ollama-setup link.
- Launch via the Cognithor Start-menu entry or
cognithorat the command prompt.
No separate Python install required.
Option B — pip install (Linux, macOS, DIY Windows)
# 1. Install Ollama (one-time)
# Linux: curl -fsSL https://ollama.com/install.sh | sh
# macOS: brew install ollama
# Win: installer from https://ollama.com
# 2. Pull the Ollama model
ollama pull qwen3:8b
# 3. Install Cognithor
pip install "cognithor[all]"
# 4. Verify
cognithor --version
# → Cognithor 0.93.0
Virtual environment recommended:
python -m venv .venv
source .venv/bin/activate # Linux/macOS
# or
.venv\Scripts\activate # Windows
pip install "cognithor[all]"
Option C — Docker Compose (headless / server)
git clone https://github.com/Alex8791-cyber/cognithor.git
cd cognithor
docker compose up -d
# → Ollama + Cognithor API on http://localhost:8741
Health check:
curl http://localhost:8741/health
# → {"status":"ok","version":"0.93.0"}
Verification
All three options should eventually answer these commands:
cognithor --version
# → Cognithor 0.93.0
ollama list
# → NAME ID SIZE MODIFIED
# → qwen3:8b abc123def 4.7 GB ...
cognithor tools list | head -5
# → Lists the first five available MCP tools
Errors? → See 07 · Troubleshooting.
Next: 01 · First Crew