The Agent Volumes Organization develops open specifications for packaging, distributing, and verifying AI agent components across runtimes, registries, package managers, and developer tools.
Agent Volumes is a Founder-led, TSC-first open specification project. The specification is vendor-neutral and implementation-agnostic: no runtime, registry, hosted service, vendor, or downstream implementation defines the standard.
| Goal | Link |
|---|---|
| Read the specification | Agent Volumes Specification |
| Build a compatible implementation | Implementers Guide |
| List your organization as adopter | ADOPTERS.md |
| Propose or discuss spec changes | Spec issues and discussions |
| Review organization governance | Charter and Governance |
| Project | Description | Status |
|---|---|---|
| Agent Volumes Specification | Packaging, distribution, and verification standard for AI agent components | Release Candidate |
Agent Volumes is led by Yunseo Kim as Founder and Project Lead, with a provisional Technical Steering Committee (TSC) as the primary deliberative and technical review body.
- The Charter defines mission, scope, conformance boundaries, and Founder/TSC authority.
- The Governance document explains roles, decision flow, Founder approval, TSC review, maintainers, Working Groups, and amendments.
- Governance proposals and TSC participation questions belong in
agent-volumes/.github.
- Contributors: review the contribution guidelines and open issues or pull requests in the repository that owns the artifact.
- Implementers: use the Implementers Guide when building compatible runtimes, clients, registries, validation tools, or packaging workflows.
- Adopters and evaluators: if your organization uses or is evaluating Agent Volumes, consider listing it in ADOPTERS.md.
- Organizations interested in deeper participation: open a governance discussion or issue about TSC participation, Working Groups, implementation feedback, or future member-company participation.
All specifications and code are released under the Apache License 2.0 unless otherwise noted.
