oasf  by agntcy

A schema framework for defining and managing AI agent capabilities

Created 11 months ago
261 stars

Top 97.5% on SourcePulse

GitHubView on GitHub
Project Summary

The Open Agentic Schema Framework (OASF) provides a standardized schema system for defining and managing AI agent capabilities, interactions, and metadata. It targets developers and researchers building interoperable AI agent systems, offering a structured approach to agent definition and discovery through attribute-based taxonomies. OASF enables consistent representation and management of agent attributes, skills, and relationships, facilitating cross-system compatibility.

How It Works

OASF employs an attribute-based taxonomy inspired by the Open Cybersecurity Schema Framework (OCSF). Its core is the record object, a primary data structure for agentic AI metadata, which can be annotated with skills and domains for discovery. Information is extended modularly via modules. This approach facilitates standardization, validation, and interoperability across distributed agent systems. Schema versions are immutable post-release, ensuring backward compatibility while allowing framework evolution.

Quick Start & Requirements

Development is managed via a Taskfile and containerized environment.

  • Primary Install/Run: Clone the repo, then use task deps, task build, task build:images, and task up for local deployment. task down handles cleanup. task reload enables schema hot-reloading.
  • Prerequisites: Taskfile, Docker (with Buildx), Go, yq, curl, tar.
  • Documentation: Links to the contributing guide, server guide, Agent Record guide, Taxonomy of AI Agent Skills, and MCP Server README are referenced within the repository.
  • Hosted Services: OASF Schema Server is available at schema.oasf.outshift.com. Protocol Buffer definitions are at https://buf.build/agntcy/oasf.

Highlighted Details

  • Schema Discovery: Browse schema versions, domains, and skills taxonomy.
  • Record Generation: Tools and prompts (via Directory MCP Server + LLM) for automatic OASF record creation from codebases.
  • Validation: Schema validation with iterative error correction capabilities.
  • Format Conversion: Import MCP/A2A formats and export with automatic enrichment.
  • Hot Reload: Supports live schema updates during development.
  • Extensibility: Allows private schema extensions and modular additions.

Maintenance & Community

The project is maintained by AGNTCY (https://github.com/agntcy). No specific community channels (like Discord/Slack) or roadmap links are detailed in the provided text.

Licensing & Compatibility

Distributed under the Apache 2.0 License. This license is permissive and generally compatible with commercial use and closed-source linking.

Limitations & Caveats

Extending records with arbitrary JSON objects in the modules list is discouraged due to skipped validation. Local setup requires Docker and a Kind cluster, potentially increasing initial setup complexity. Backend changes necessitate re-running the build and deployment commands.

Health Check
Last Commit

2 days ago

Responsiveness

Inactive

Pull Requests (30d)
7
Issues (30d)
6
Star History
6 stars in the last 30 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
7 more.

SuperAGI by TransformerOptimus

0.2%
17k
Open-source framework for autonomous AI agent development
Created 2 years ago
Updated 11 months ago
Feedback? Help us improve.