OpenAI and Anthropic donate the Open Responses API specification and the Agent Skills open standard to the Linux Foundation’s Agentic AI Foundation, as the foundation expands its hosted open-source AI portfolio.

Decision brief

What changed

OpenAI donated its Open Responses API specification and related frameworks to the Linux Foundation’s Agentic AI Foundation, and Anthropic donated the Agent Skills open standard to the same foundation. Both moves shift these agent-related specifications into a vendor-neutral governance home while keeping existing community GitHub resources and branding accessible.

Why it matters

For leaders building or buying AI agents, these donations increase the odds that key integration layers for model calls and reusable agent capabilities will be governed outside any single vendor, which can reduce dependence on proprietary interfaces. If the standards gain broader multi-provider support as described in the coverage, enterprises may face lower integration and switching costs when connecting agents across model providers and platforms. The decision relevance is highest for roadmap, platform, and partner choices rather than near-term revenue assumptions.

Affected roles

CEO COO CTO

Evidence

The New Stack reported OpenAI’s donation of Open Responses and framed it as a way to create shared LLM API calls across providers. Simon Willison and Anthropic’s own announcement both reported Anthropic’s donation of Agent Skills to AAIF and consistently described AAIF as a more open, community-driven, vendor-neutral governance home; Anthropic also provided adoption figures for Agent Skills.

What remains uncertain

The coverage does not establish whether major providers beyond the donors will implement these standards, how quickly enterprise tooling will align, or how AAIF governance will operate in practice. Reported adoption metrics for Agent Skills come from Anthropic’s own announcement, and the practical effect on interoperability, migration cost, and vendor leverage remains an assumption until broader independent uptake is visible.

Monitor next

Watch for additional major model providers and agent platform vendors publicly committing to implement or govern Open Responses and Agent Skills under AAIF.

Analytical support, not advice — assumptions and open questions stated above.

Source coverage