The canonical home for all Works With Agents (WWA) agent interoperability specifications.
This repository contains the complete set of open specifications that define how AI agents discover, communicate, transact, coordinate, and trust each other. All specs are organized around the Agent OSI Model — a 7-layer framework that gives the agent ecosystem a shared vocabulary for building, debugging, and composing agent infrastructure.
Browse all specs with rich formatting and cross-references at:
👉 workswithagents.dev/specs/
This repo has agent-friendly issues — structured tasks with clear input, output, and success criteria, designed for AI agent execution.
| Issue | Description | Tags |
|---|---|---|
| #46 | Script: Handoff state viewer (CLI) | agent-friendly, community |
| #45 | Script: llms.txt generator for WWA specs | agent-friendly, community |
| #44 | Setup: Conformance check in CI | agent-friendly, community |
| #43 | Script: Registry heartbeat monitor | agent-friendly, community |
| #42 | Script: Deploy two handoff-capable reference agents | agent-friendly, community |
| #40 | Reference Agent: MCP server wrapper | help-wanted, hacktoberfest |
| #39 | Reference Agent: Kubernetes deployment manifest | help-wanted, hacktoberfest |
| #38 | Conformance Suite: Agent Registry protocol tests | help-wanted, hacktoberfest |
| #37 | Conformance Suite: Identity Protocol tests | help-wanted, hacktoberfest |
| #36 | Reference Agent: Rust implementation | help-wanted, hacktoberfest |
Agents: filter by label agent-friendly or help-wanted to find ready-to-execute issues.
Humans: see CONTRIBUTING.md for contribution guidelines.
┌─────────────────────────────────────────────────────┐
│ Layer 7 — Governance & Audit │
│ Compliance, SLAs, transactions, attestation, │
│ economics, payments, security disclosure │
├─────────────────────────────────────────────────────┤
│ Layer 6 — Presentation / Payload │
│ MCP UI payload schemas, message formatting │
├─────────────────────────────────────────────────────┤
│ Layer 5 — Coordination │
│ IACP, ACP, ECP, delegation, fault tolerance │
├─────────────────────────────────────────────────────┤
│ Layer 4 — Session │
│ Handoff, clarification, context management │
├─────────────────────────────────────────────────────┤
│ Layer 3 — Discovery │
│ Capability manifests, agent registry, trust scores │
├─────────────────────────────────────────────────────┤
│ Layer 2 — Identity & Auth │
│ Agent identity, onboarding, DID-based auth │
├─────────────────────────────────────────────────────┤
│ Layer 1 — Runtime / Infrastructure │
│ Deployment manifests, local-first, execution env │
└─────────────────────────────────────────────────────┘
| Spec | Version | Status | Description |
|---|---|---|---|
| Agent OSI Model | 1.0.0 | Published | 7-layer reference architecture for agent infrastructure |
| Spec | Version | Status | Description |
|---|---|---|---|
| Deployment Manifest | 1.0.0 | Published | Declarative agent deployment specification |
| Local-First Certification | 1.0.0 | Published | Requirements for agents operating offline-first |
| Spec | Version | Status | Description |
|---|---|---|---|
| Onboarding Protocol | 1.0.0 | Published | Agent registration and initial trust establishment (L1/L3) |
| Identity Protocol | 1.0.0 | Published | Agent identity, authentication, and DID-based auth (L2/L3) |
| Spec | Version | Status | Description |
|---|---|---|---|
| Agent Capability Manifest | 1.0.0 | Published | Machine-readable agent capability declarations |
| Agent Registry | 1.0.0 | Published | Registry protocol for agent discovery and lookup (L3/L7) |
| Trust Score | 1.0.0 | Published | Computational trust scoring model (L3/L5) |
| Spec | Version | Status | Description |
|---|---|---|---|
| Handoff Protocol | 1.0.0 | Published | Structured agent-to-agent session handoff |
| Clarification Protocol | 1.0.0 | Published | Requesting and resolving ambiguity between agents |
| Spec | Version | Status | Description |
|---|---|---|---|
| Coordination Protocol (ACP) | 1.0.0 | Published | Agent coordination and task orchestration |
| IACP | 1.0.0 | Published | Inter-Agent Communication Protocol |
| IACP Internet Draft | draft-00 | Draft | IACP in IETF RFC format |
| IACP Fault Tolerance | 1.0.0 | Draft | Dead-letter and rollback protocol for IACP |
| Ephemeral Communication Protocol (ECP) | 1.0.0 | Published | Short-lived, TTL-bound agent communication |
| Delegation Framework | 1.0.0 | Published | Agent task and authority delegation (L5/L7) |
| Spec | Version | Status | Description |
|---|---|---|---|
| Transaction Protocol (ATP) | 1.0.0 | Published | Agent-to-agent commerce and transaction ledger |
| SLA Framework | 1.0.0 | Published | Agent service-level agreement specification |
| Compliance-as-Code | 1.0.0 | Published | Machine-enforceable compliance rules |
| Security Disclosure Protocol | 1.0.1 | Published | Coordinated vulnerability disclosure for agents |
| Agent Economics Protocol | 1.0.0 | Published | Agent pricing, billing, and economic models |
| AP2 — Payment Mandate | 0.9.0 | Published | Agent spending authority and payment execution |
| Attestation Protocol | 1.0.0 | Published | Cryptographic attestations and claims |
| Auditor Verification | 1.0.0 | Published | Third-party agent audit and verification |
| Reputation Ledger | 1.0.0 | Published | Decentralized agent reputation tracking |
| Spec | Version | Status | Description |
|---|---|---|---|
| ASFS — Agent Skill Format Standard | 1.0.0 | Published | Universal skill packaging format |
| AI Gateway / PEP | 1.0.0 | Published | Policy enforcement point for agent traffic |
| Reputation Ledger | 1.0.0 | Published | Cross-layer reputation infrastructure |
| Agent Coding Benchmark | 1.0.0 | Published | Standardized agent coding capability assessment |
Every spec follows a consistent structure:
- YAML frontmatter —
id,title,version,status,authors,date - Status — lifecycle marker (Draft, Review, Published, Deprecated)
- Abstract — one-paragraph summary
- Motivation — why this spec exists
- Specification — detailed content using RFC 2119 conventions (MUST, SHOULD, MAY)
- Security Considerations — security implications
- References — related specs and standards
- For new specs: Open a "New Spec" issue describing the problem and proposed solution.
- For existing specs: Open a "Spec Change" issue specifying the section and rationale.
- See CONTRIBUTING.md for the full process.
GET https://workswithagents.dev/specs/index.md → All specs
GET https://workswithagents.dev/specs/{spec}.md → Individual spec
- Python:
pip install workswithagents - TypeScript:
npm install @workswithagents/agent-foundry - Source: github.com/workswithagents/works-with-agents
specs/
├── README.md ← You are here
├── CONTRIBUTING.md ← How to contribute
├── LICENSE ← CC BY 4.0
├── GOVERNANCE.md ← Spec governance and decision-making
├── spec-template.md ← Template for new spec proposals
├── index.md ← Machine-readable spec index
├── *.md ← Individual specification files (28 specs)
├── agent-flags.json ← Agent capability flags reference
├── mcp-ui-payload-schema.json ← MCP UI payload schema
└── .github/
└── ISSUE_TEMPLATE/
├── new-spec.md ← Template for proposing new specs
└── spec-change.md ← Template for spec change requests
Specifications are maintained by the Works With Agents project. See GOVERNANCE.md for:
- Decision-making process (consensus-based)
- Versioning scheme (MAJOR.MINOR.PATCH)
- Deprecation policy
- How to become a maintainer
All specifications in this repository are licensed under the Creative Commons Attribution 4.0 International (CC BY 4.0) license.
You are free to:
- Share — copy and redistribute the material in any medium or format
- Adapt — remix, transform, and build upon the material for any purpose, even commercially
Under the following terms:
- Attribution — you must give appropriate credit, provide a link to the license, and indicate if changes were made
See LICENSE for the full legal text.
28 specifications — actively maintained. New proposals welcome via issues.