Skip to content

ADL v0.1.0 — Introducing the Agent Definition Language: A Community-Driven Standard for AI Agents

Latest
Latest

Choose a tag to compare

View all tags
@swanandrao swanandrao released this 04 Dec 22:54
open-standard
03e258b
This commit was created on GitHub.com and signed with GitHub’s verified signature.
GPG key ID: B5690EEEBB952194
Verified
Learn about vigilant mode.

**ADL v0.1.0 — Introducing the Agent Definition Language

A Community-Driven, Model-Agnostic Standard for AI Agents**
We’re excited to publish the first open-source release of ADL (Agent Definition Language) a vendor-neutral, model-agnostic, framework-independent standard for defining AI agents, their roles, capabilities, tools, schemas, and governance rules.
ADL was born out of a simple observation:
everyone is defining agents differently, in parallel, with no shared language or interoperability.
This release is an invitation to the developer community to help shape a common foundation for agent development.

🚀 What’s Included in v0.1.0

  • Core Specification
  • Formal JSON Schema for agent definitions

  • Formal JSON Schema for tool/function definitions

  • Governance primitives: constraints, audit metadata, safety blocks

  • Reference validation logic and utilities

Examples

  • Sample ADL agent definitions (simple → advanced)

  • Tool examples including parameter schemas and execution metadata

  • Example agent capabilities and structured workflows
  • Documentation
  • Design principles and goals

  • Rationale behind ADL

  • Intended future extensions

  • Comparison with other agent frameworks

🔑 Why ADL Matters

  • ADL aims to reduce fragmentation and standardize how agents are described across the ecosystem.
  • Key Advantages
  • Model-agnostic
Works with GPT, Claude, Gemini, Mistral, Llama.cpp, and any future model.
  • Framework-neutral
Usable with LangChain, LlamaIndex, AutoGen, custom orchestrators, or any runtime.
  • Composable & deterministic
Clear schema definitions lead to predictable agent execution.
  • Portable & interoperable
Share agent definitions across teams, platforms, and environments without rewriting.
  • Community-first
The spec evolves through open contributions, not locked behind proprietary systems.

🧱 Design Principles
Simplicity - A clean, minimal, readable structure

Extensibility - Built to expand as the ecosystem evolves

Predictability - Schema-driven, strict, and unambiguous

Interoperability - Works across tools, languages, and infrastructures

Transparency - A standard that belongs to the community

🗺️ Roadmap
v0.2 - Memory schema

v0.3 - Workflow schema (multi-step, multi-agent)

v0.4 - Tool categories + integration metadata

v1.0 - Community-approved stable standard
Roadmap feedback is welcome the spec is intentionally early to invite discussion.

🤝 How to Get Involved
We’d love contributions from developers, researchers, framework authors, and platform builders.
Ways to contribute: - Review the spec and open issues
- Propose extensions for memory, safety, workflows, or evaluations
- Add examples, templates, or validation tooling
- Build adapters for LangChain, LlamaIndex, AutoGen, or custom runtimes
- Participate in GitHub Discussions
ADL will only succeed through community involvement.

📎 Links
Deep-Dive Blog: https://www.nextmoca.com/blogs/agent-definition-language-adl-the-open-source-standard-for-defining-ai-agents

Assets 2
Loading