agents-json  by wild-card-ai

Open specification to translate OpenAPI into LLM tools

created 6 months ago
1,264 stars

Top 32.0% on sourcepulse

GitHubView on GitHub
Project Summary

This project defines the agents.json specification, an open standard for describing API and agent interactions, built upon OpenAPI. It aims to simplify how AI agents discover and execute API calls by providing structured contracts, enabling developers to integrate AI agents with existing web services more reliably and efficiently.

How It Works

The agents.json specification extends OpenAPI by adding constructs like "flows" and "links" to define sequences of API calls and their interdependencies. This allows for richer descriptions of API capabilities tailored for LLM consumption, including optimized descriptions and examples. The Wildcard Bridge Python package parses these specifications to orchestrate API calls based on agent requests, supporting various authentication methods.

Quick Start & Requirements

  • Install Wildcard Bridge: pip install wildcard-bridge
  • Requires Python 3.7+.
  • Examples are provided as Jupyter notebooks (.ipynb) in the ./examples/ directory, demonstrating integration with services like Resend, Stripe, and Google Sheets.
  • Official documentation and demos are linked within the README.

Highlighted Details

  • Builds on the OpenAPI standard, leveraging existing API descriptions.
  • Introduces "flows" and "links" to define multi-step API call sequences for AI agents.
  • Designed for LLM consumption, optimizing descriptions and argument generation.
  • Supports Basic, ApiKey, and Bearer authentication.

Maintenance & Community

  • Project initiated by Wildcard AI.
  • Community feedback is encouraged via their Discord server.
  • Active social media presence on Twitter.

Licensing & Compatibility

  • agents.json specification: Apache 2.0 License.
  • Wildcard Bridge package: Affero GPL v3. This strong copyleft license may restrict commercial use or linking with closed-source applications.

Limitations & Caveats

The project is in its early stages (specification version 0.1.0) with several features planned for the roadmap, including memory management, runtime field transformations, rate-limiting, and error handling. The Affero GPL v3 license for the Wildcard Bridge package requires careful consideration for commercial adoption.

Health Check
Last commit

2 months ago

Responsiveness

1 day

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

Explore Similar Projects

Starred by Elie Bursztein Elie Bursztein(Cybersecurity Lead at Google DeepMind), Chip Huyen Chip Huyen(Author of AI Engineering, Designing Machine Learning Systems), and
3 more.

composio by ComposioHQ

0.2%
26k
Toolset for AI agent integrations via function calling
created 1 year ago
updated 22 hours ago
Feedback? Help us improve.