Memento-Skills

Making Prompts Stateful and Agents Continually Learnable

Framework

Memento-Skill framework diagram

Option A: Scripted Install on macOS (Recommended)

Use this for the fastest setup on macOS.

  • Prerequisites: git, curl, optional tmux
git clone https://github.com/HuichiZhou/Memento-S.git
cd Memento-S
./install_mac.sh

Option B: Manual Install on macOS or Linux

Use this if you want full control over environment setup.

1) Clone source

git clone https://github.com/HuichiZhou/Memento-S.git
cd Memento-S

2) Create venv and install dependencies

python3 -m venv .venv
source .venv/bin/activate
pip install -U pip
pip install -r requirements.txt

3) Configure LLM

.venv/bin/python cli/main.py config
LLM_API=openai
LLM_MODEL=gpt-4o
LLM_API_KEY=your-api-key
LLM_BASE_URL=https://api.openai.com/v1
LLM_MAX_TOKENS=4096

4) Run

.venv/bin/python tui.py run
.venv/bin/python cli/main.py agent
.venv/bin/python cli/main.py agent -m "Your prompt"

Option C: Docker Workflow (If Dockerfile Exists)

docker build -t memento-s:latest .

docker run --rm -it \
  --name memento-s \
  --env-file .env \
  -p 8080:8080 \
  memento-s:latest

Post-Install Checks

.venv/bin/python cli/main.py config list
.venv/bin/python cli/main.py agent -m "hello"

Team Members