Skip to main content
v0.5.0 · Protocol-Agnostic, Platform-Independent

The Agent
Contract Specification

The normative contract for AI agents. Define identity, capabilities, trust boundaries, and governance — before a single agent runs.

View SchemaTry Validation

v0.5.0

Stable

March 2026 · Latest release

View Schema

Why This Specification?

The Problem: Four major survey papers catalog 15+ agent communication protocols — but identify zero standards for agent contracts, trust boundaries, or governance. MCP connects tools. A2A connects agents. Neither defines what an agent is, what it may do, or who is accountable.
The Solution: OSSA is the contract layer that sits above protocols and below applications. A vendor-neutral YAML specification that defines agent identity, capabilities, trust boundaries, and governance — like OpenAPI standardized REST APIs, OSSA standardizes agent contracts.
The Result: Define your agent once in OSSA. Export to any protocol (MCP, A2A) or platform (LangChain, CrewAI, Kubernetes, GitLab). No vendor lock-in. True portability. Auditable governance from day one.

Specification Structure

The OSSA specification is organized into four key areas.

1

Core Specification

The fundamental structure of an OSSA agent manifest, including metadata, specification, and extensions.

Read Core Specification
2

Schema Reference

Complete JSON Schema definition with all properties, types, constraints, and validation rules.

Explore Schema
3

Platform Extensions

Framework-specific extensions for Cursor, OpenAI, CrewAI, LangChain, Anthropic, Langflow, AutoGen, and more.

View Extensions
4

Validation Rules

Comprehensive validation rules, error messages, and best practices for creating valid agent manifests.

Try Validation

Key Concepts

Agent Manifest

A YAML or JSON file that describes an AI agent, including its role, LLM configuration, tools, and deployment requirements.

apiVersion: ossa/v0.5.0
kind: Agent
metadata:
  name: my-agent
spec:
  role: You are a helpful assistant
  llm:
    provider: openai
    model: gpt-4

Federal Governance Alignment

OSSA v0.4.9 introduces explicit support for regulated-environment agent identity and authorization, including Cedar policy integration and cryptographic provenance tracking.

Platform Extensions

Framework-specific extensions (CrewAI, LangChain, Agentscope) allow agents to leverage platform features while maintaining root OSSA contract compliance.

Normative Validation

Strict JSON Schema enforcement (ossa-0.4.9.schema.json) ensures that every agent manifest is verifiable, auditable, and production-ready.

Version History

VersionDateAPInpm
v0.5.0
latest
—ossa/v0.5.0npm →
v0.4.9—ossa/v0.4.9npm →
v0.4.8—ossa/v0.4.8npm →
v0.4.6—ossa/v0.4.6npm →
v0.4.5—ossa/v0.4.5npm →
v0.4.4—ossa/v0.4.4npm →
v0.4.3—ossa/v0.4.3npm →
v0.4.2—ossa/v0.4.2npm →
v0.4.1—ossa/v0.4.1npm →
v0.4.0—ossa/v0.4.0npm →
v0.3.6—ossa/v0.3.6npm →
v0.3.5—ossa/v0.3.5npm →
v0.3.3—ossa/v0.3.3npm →
v0.3.2—ossa/v0.3.2npm →
v0.3.1—ossa/v0.3.1npm →
v0.3.0—ossa/v0.3.0npm →
v0.2.9—ossa/v0.2.9npm →
v0.2.8—ossa/v0.2.8npm →
v0.2.7—ossa/v0.2.7npm →
v0.2.5-RC
RC
—ossa/v0.2.5-RCnpm →
v0.2.4—ossa/v0.2.4npm →
v0.1.9—ossa/v0.1.9npm →

View all versions on npm

Resources

Interactive Schema Explorer

Explore the complete JSON Schema interactively.

Agent Builder

Test and validate your agent manifests in real-time.

Examples Gallery

Browse real-world examples and use cases.

Full Documentation

Comprehensive guides and tutorials.

Start Building with OSSA

Define your agents once, deploy everywhere. Join the open standard for portable AI agents.

Get StartedJoin Discord