tempLLM

local llm api — no sdk, no api key

spin up a local server · hit clean REST endpoints · get responses

how it works

templlm drives a live browser session via Playwright and exposes it as a FastAPI server with REST endpoints. the templlm CLI talks to that server over HTTP.

templlm "prompt"  →  POST /ask  →  FastAPI server  →  Playwright  →  ChatGPT  →  response

requirements

| dependency | version | notes | |------------|---------|-------| | Node.js | 18+ | runs the CLI | | Python | 3.8+ | runs the server | | Google Chrome | any recent | for authenticated mode (recommended) |

install

option A — npm (recommended)

npm install -g templlm
templlm init        # interactive setup wizard

option B — clone the repo

git clone https://github.com/YOUR_USER/tempLLM.git
cd tempLLM
npm install         # installs pip packages + playwright chromium automatically
npm install -g .    # puts `templlm` in your PATH
templlm init        # interactive setup wizard

setup wizard

templlm init detects your OS and walks you through everything:

┌────────────────────────────────────────┐
│          templlm  ·  setup wizard       │
└────────────────────────────────────────┘

Detected OS:  Linux (Arch Linux)
Python:       ✓ 3.11.0  (python3)

Which connection mode?
  1  Mode A — CDP  (recommended)  Connect to your own Chrome with an active session
  2  Mode B — Headless            Playwright launches Chromium in the background

›

It then prints the exact Chrome launch command for your OS, writes .env automatically, and optionally runs the login flow.

platform setup

# Arch / Manjaro
sudo pacman -S python python-pip nodejs npm
yay -S google-chrome          # for Mode A (CDP)

# Ubuntu / Debian
sudo apt install python3 python3-pip nodejs npm
# Chrome: https://google.com/chrome

# Fedora
sudo dnf install python3 python3-pip nodejs npm
sudo dnf install google-chrome-stable

npm install -g templlm
templlm init

PATH note: npm global bins land in ~/.local/bin. Make sure it's in your PATH:

echo 'export PATH="$HOME/.local/bin:$PATH"' >> ~/.bashrc   # bash
echo 'export PATH="$HOME/.local/bin:$PATH"' >> ~/.zshrc    # zsh
source ~/.bashrc   # or open a new terminal

# Install prerequisites via Homebrew
brew install python node
brew install --cask google-chrome    # for Mode A (CDP)

npm install -g templlm
templlm init

# Install prerequisites via winget
winget install Python.Python.3
winget install OpenJS.NodeJS
winget install Google.Chrome        # for Mode A (CDP)

npm install -g templlm
templlm init

Or use Chocolatey:

choco install python nodejs googlechrome
npm install -g templlm
templlm init

Note: Run PowerShell as Administrator for global npm installs, or configure a user-local npm prefix:

npm config set prefix "$env:APPDATA\npm"
# add %APPDATA%\npm to your PATH in System Environment Variables

usage

templlm init                          # first-time setup
templlm "give me a bubble sort"       # single response
templlm --stream "explain async/await"  # streaming response
templlm --setup                       # re-run login (session expired)

The server starts automatically in the background when you run a prompt. No need to run python run.py manually.

connection modes

mode A — CDP (recommended)

Connect to your existing Chrome with a live logged-in session.

1. Launch Chrome with remote debugging:

google-chrome-stable --remote-debugging-port=9222 --user-data-dir=/tmp/chrome-cdp-profile

"/Applications/Google Chrome.app/Contents/MacOS/Google Chrome" \
  --remote-debugging-port=9222 --user-data-dir=/tmp/chrome-cdp-profile

& "C:\Program Files\Google\Chrome\Application\chrome.exe" `
  --remote-debugging-port=9222 `
  --user-data-dir=C:\temp\chrome-cdp-profile

2. Log in to ChatGPT in that window.

3. Run templlm init or set CDP_URL=http://localhost:9222 in .env.

The Chrome profile is saved to --user-data-dir — you only log in once.

mode B — headless

Don't open Chrome. Just run templlm "prompt" and Playwright handles its own Chromium. Session is limited to unauthenticated access unless you have a saved session.json.

endpoints

| method | endpoint | description | |--------|----------|-------------| | POST | /ask | full JSON response | | POST | /ask/stream | server-sent events (SSE) stream | | GET | /health | server & browser status | | POST | /screenshot | debug screenshot, returns path | | POST | /session/invalidate | clear saved session |

Interactive docs → http://localhost:8000/docs

test the api

curl

# Linux / macOS
curl -X POST http://127.0.0.1:8000/ask \
  -H "Content-Type: application/json" \
  -d '{"prompt": "give me a bubble sort"}'

# Windows PowerShell
Invoke-RestMethod -Method Post -Uri http://127.0.0.1:8000/ask `
  -ContentType "application/json" `
  -Body '{"prompt": "give me a bubble sort"}'

python test client (no extra installs)

python test_client.py "give me a bubble sort"
python test_client.py --stream "give me a bubble sort"

configuration

# .env

CDP_URL=http://localhost:9222   # blank = Mode B (headless)

HEADLESS=false                  # true = no visible browser in Mode B
SESSION_FILE=./session.json
SLOW_MO=0
RESPONSE_TIMEOUT=120

HOST=0.0.0.0
PORT=8000

project structure

tempLLM/
│
├── app/
│   ├── main.py          FastAPI app + lifespan
│   ├── config.py        env config
│   ├── models.py        pydantic schemas
│   ├── browser.py       Playwright automation + mode detection
│   └── routes/
│       └── ask.py       endpoints
│
├── bin/
│   └── cli.js           npm CLI entry point
│
├── scripts/
│   ├── init.js          interactive setup wizard
│   └── postinstall.js   runs on npm install
│
├── run.py               server entry point
├── cli.py               direct browser CLI (no server)
├── test_client.py       HTTP test client
├── requirements.txt     Python dependencies
└── package.json         npm package