Integuru is an open-source Python tool that reverse-engineers browser network requests to generate runnable integration code for platforms without official APIs. It builds a dependency graph of requests and can generate code to perform actions via internal endpoints.
Collecting history — the radar snapshots this repo daily. The trend line appears after 3 days of data (1 so far).
What it is
Integuru is a Python-based agent that generates runnable integration code by reverse-engineering browser network requests to hit a platform's internal endpoints. It outputs code and maintains a dependency graph to orchestrate requests.
How it works
- Uses browser network requests to create a runnable Python file that hits a platform's internal endpoints.
- Identifies dynamic parts within a request (e.g., accountId, userId) and finds requests that provide these parts.
- Builds a dependency graph by attaching dependent requests to the original request and traversing toward a master node.
- Repeats until a node depends only on authentication cookies.
- Converts each node in the graph to a runnable function as it traverses from nodes with no outgoing edges to the master node.
Getting started
-
Setup OpenAI API keys and set OPENAI_API_KEY environment variable.
-
Install Python requirements via poetry:
poetry install -
Open a poetry shell:
poetry shell -
Register the Poetry virtual environment with Jupyter:
poetry run ipython kernel install --user --name=integuru -
Spawn a browser to generate HAR data:
poetry run python create_har.pyLog into the platform and perform the desired action.
-
Run Integuru:
poetry run integuru --prompt "download utility bills" --model <gpt-4o|o3-mini|o1|o1-mini>You can also run via Jupyter Notebook main.ipynb
Recommended to use gpt-4o as the model for graph generation as it supports function calling. Integuru will automatically switch to o1-preview for code generation if available in the user's OpenAI account.
-
Usage: after setup, provide the HAR file and a prompt describing the action.
poetry run integuru --helpOptions include model, prompt, har-path, cookie-path, max_steps, input_variables, generate-code, and help.
Recent releases
- There are no releases listed in this repository metadata.
Traction
- Stars: 4653
Behind the repo
- License: AGPL-3.0
- Language: Python
- Created: 2024-10-22
- Last push: 2026-06-24
- Repository topics include agent, ai-agent, api, automation, integration, llm, open-source, rpa, unofficial-api
Caveats
- License: AGPL-3.0
- There are 24 open issues.
- No releases are listed in the latest releases section.