Repolode
Submit
agntcy

agntcy/oasf

A standardized JSON schema framework and server for describing AI agent capabilities, skills, domains and metadata so agents can be validated and discovered.

View on GitHubWebsite
Stars
340
Forks
47
Open issues
19
Last commit
yesterday

What is it?

OASF (Open Agentic Schema Framework) defines a common data structure for AI agent records. Each record is described with attribute-based taxonomies of skills and domains, plus optional modules that extend records in a composable way. The goal is consistent definition, validation and discovery of agent capabilities across distributed systems.

The project is modeled after the OCSF (Open Cybersecurity Schema Framework) in its data modeling approach, and its schema server is a derivative of the OCSF schema server. It ships an HTTP schema server for browsing and validating schemas, along with Protocol Buffer definitions published to the Buf Schema Registry.

Development is handled through a Taskfile-based workflow with a containerized environment, including a local Kind cluster deployment via Helm and a hot-reload mode for schema changes. Private schema extensions are supported, and an MCP server can help LLMs generate and validate records.

Key features

  • 01Defines a standard record structure for describing AI agents.
  • 02Annotates agents with skills and domains taken from attribute-based taxonomies.
  • 03Supports private schema extensions with validation.
  • 04Provides an HTTP schema server for browsing and validating schemas.
  • 05Includes hot-reload during schema development and Helm-based local deployment.
  • 06Publishes Protocol Buffer definitions to the Buf Schema Registry.

Use cases

  • Standardizing how agent capabilities are described for discovery across systems.
  • Validating agent records against a shared schema during development.
  • Hosting a private schema server with your own extensions.
  • Letting an LLM generate and validate OASF records via an MCP server.

Quick start

  1. 1

    Clone the repository

    git clone https://github.com/agntcy/oasf.git
  2. 2

    Install the prerequisites (Taskfile, Docker with Buildx, Go, yq, curl, tar)

  3. 3

    Build the server image locally

    IMAGE_TAG=latest task build:server
  4. 4

    Deploy OASF locally on an ephemeral Kind cluster, then open localhost:8080

    task up

Things to know

  • Local deployment requires Docker with Buildx, Go, yq and a Taskfile setup.
  • Changes to the server backend require re-running task up, so hot reload only covers schema changes.

Topics

#ai-agents#schema#json-schema#taxonomy#validation#protocol-buffers#interoperability

FAQ

What is oasf?
A standardized JSON schema framework and server for describing AI agent capabilities, skills, domains and metadata so agents can be validated and discovered. OASF (Open Agentic Schema Framework) defines a common data structure for AI agent records. Each record is described with attribute-based taxonomies of skills and domains, plus optional modules that extend records in a composable way. The goal is consistent definition, validation and discovery of agent capabilities across distributed systems.
Is oasf free to use? What license does it have?
oasf is open source under the Apache-2.0 license. Check the license terms before using it in commercial products.
Is oasf still maintained?
Actively maintained. The last commit to oasf was on Oct 21, 2026.
What can I use oasf for?
Standardizing how agent capabilities are described for discovery across systems. Validating agent records against a shared schema during development. Hosting a private schema server with your own extensions. Letting an LLM generate and validate OASF records via an MCP server.
How do I install oasf?
Clone the repository: git clone https://github.com/agntcy/oasf.git Install the prerequisites (Taskfile, Docker with Buildx, Go, yq, curl, tar) Build the server image locally: IMAGE_TAG=latest task build:server Deploy OASF locally on an ephemeral Kind cluster, then open localhost:8080: task up
Where can I get oasf?
The source code and installation instructions are on GitHub: https://github.com/agntcy/oasf

Summary generated by AI from the project's README. GitHub stats refresh daily.

Related projects