Web Agent Protocol (WAP) provides a Python SDK to record and replay browser interactions, convert to MCP servers, and run replays via WAP-Replay protocol. Latest release introduces a desktop tool for exact/smart replays.
Collecting history — the radar snapshots this repo daily. The trend line appears after 3 days of data (2 so far).
What it is
The Web Agent Protocol (WAP) is a Python SDK that records browser interactions, converts the raw event stream into exact‑replay or smart‑replay action lists, converts recorded actions into MCP servers, and replays those lists using the WAP-Replay protocol. It supports collecting data via an OTA‑WAP Chrome extension and provides tooling to generate replay lists and run replays.
How it works
- Collect user interaction data with the OTA‑WAP Chrome extension and a local data server (action data saved under data/YYYYMMDD/taskid/summary_event_<timestamp>.json).
- Convert event streams into either exact_replay or smart_replay action lists.
- Convert recorded actions into MCP servers for reuse by agents or users.
- Replay lists with the WAP-Replay protocol via run_replay or through MCP playback.
Getting started
Install dependencies:
"pip install -r requirements.txt"
Create a conda environment:
conda create -n WAP python=3.11
Activate the environment:
conda activate WAP
Set Python path:
set PYTHONPATH=C:/path/to/webagentprotocol # for Windows
export PYTHONPATH=/path/to/webagentprotocol # for Linux
Create a .env with API keys:
OPENAI_API_KEY=sk-proj-...
DEEPSEEK_API_KEY=sk-...
Start data collection server:
python action_collect_server.py
Server URL: http://localhost:4934/action-data Sessions saved to:
data/YYYYMMDD/taskid/summary_event_<timestamp>.json
Recent releases
Latest release (wap-replay-tool-1.0.0, 2025-06-01) is a desktop app for running WAP smart and exact replays.
This version includes:
- JSON loader & validator
- auto check
exact_replay/smart_rep