Public roadmap

CueFrame is one shared video project for people and agents: a human can work in the desktop editor while an agent reads and patches the same composition through MCP, the API, CLI, or x402. This page describes the public priorities for making that workflow useful and trustworthy.

It is a direction of travel, not a release contract. We do not attach dates to roadmap items, and priorities may change as we learn from users and contributors. Shipped work appears in the changelog.

Now: make the shared substrate easy to trust

• Keep the desktop editor, composition contracts, SDK, CLI, MCP package, and documentation legible and buildable from this repository.
• Make edits inspectable: composition changes, validation, previews, and render results should give both a human reviewer and an agent clear evidence of what changed.
• Keep the local-first desktop workflow clear: footage stays on-device until a person chooses to use a hosted capability or share a project with an agent.

Next: show the real workflows

• Publish repeatable, real recordings for the agent paths: MCP setup and composition, CLI workflows, and keyless x402 calls.
• Add recordings for media discovery workflows, including stock import and semantic library search, using clear test data and visible outcomes.
• Keep expanding the human-side demo set: transcript edits, AI Director changes, graphics, captions, review, and export.
• Grow the public template catalog past the Yosemite pair: each template is one document with its component, timeline, grade, captions, audio and review rubric, in the public CueFrame templates repository.

Next: make integration predictable

• Improve end-to-end examples for MCP, REST, CLI, SDK, webhooks, and x402 so an agent can choose an interface without learning a different video model.
• Keep generated API surfaces tied to the checked-in OpenAPI contract, with clear source paths and regeneration guidance for contributors.
• Strengthen standalone-package and workspace checks so a clean clone and a published package both behave as documented.

Longer term: a better human–agent handoff

• Reduce the friction of moving between an agent's proposed composition and a human's review, edit, and approval in the desktop app.
• Make the evidence behind a video decision—media context, composition intent, preview, validation, and render result—easier to inspect and share.
• Grow an open ecosystem of components, skills, integrations, and examples around the composition model.

Help shape it

Open a feature request with the workflow you are trying to automate or review. The most useful reports include the agent/client you use, the input media you expect to handle, the human approval point, and what would make the result safe to ship.