Integuru  by Integuru-AI

AI agent for permissionless integrations via reverse-engineered APIs

created 9 months ago
4,412 stars

Top 11.3% on sourcepulse

GitHubView on GitHub
Project Summary

Integuru is an AI agent designed to automate the creation of software integrations by reverse-engineering platform internal APIs. It targets developers and power users seeking to build custom connections to web services without official API access, generating runnable Python code for specific actions.

How It Works

Integuru analyzes network requests captured in a HAR file, identifying dynamic parameters and dependencies between requests. It constructs a dependency graph, starting from the desired action and tracing back to requests that provide necessary data (like user IDs or account IDs). This graph is then translated into Python code that executes the sequence of API calls, abstracting away the complexity of manual reverse-engineering.

Quick Start & Requirements

  • Install dependencies: poetry install
  • Set OpenAI API key: export OPENAI_API_KEY='...'
  • Generate HAR file: poetry run python create_har.py
  • Run Integuru: poetry run integuru --prompt "download utility bills"
  • Requires Python 3.12, Poetry, and OpenAI API access (GPT-4o recommended for graph generation).
  • Official Docs: main.ipynb

Highlighted Details

  • Generates Python code for platform integrations by reverse-engineering internal APIs.
  • Creates dependency graphs of network requests to reconstruct complex workflows.
  • Supports input variables for actions (e.g., specifying a year for document downloads).
  • Leverages LLMs (GPT-4o, o1-preview) for analysis and code generation.

Maintenance & Community

  • Developed by Integuru.ai.
  • Accepts custom integration requests and offers hosting/authentication services.
  • Open-source unofficial APIs available.

Licensing & Compatibility

  • License not specified in the README.
  • Compatible with Python 3.12.

Limitations & Caveats

The project is in active development, with input variable support for code generation noted as "coming soon." The effectiveness relies heavily on the quality of the captured HAR file and the clarity of the user's prompt.

Health Check
Last commit

5 months ago

Responsiveness

1 day

Pull Requests (30d)
0
Issues (30d)
0
Star History
105 stars in the last 90 days

Explore Similar Projects

Starred by Patrick von Platen Patrick von Platen(Core Contributor to Hugging Face Transformers and Diffusers) and Jeff Hammerbacher Jeff Hammerbacher(Cofounder of Cloudera).

client-python by mistralai

0.3%
628
Python SDK for Mistral AI platform
created 1 year ago
updated 1 week ago
Feedback? Help us improve.