Roadmap

What's stable, what's being added, and what's likely to affect plugin authors. Use this to decide whether your plugin can depend on something today, or whether to wait for the API to settle.

Stable The source / transform / dataset / dashboard / frontend / theme / analysis plugin types. The plugin entry-point system. The CLI installer (shenasctl source add for local wheels). Ed25519 signing for in-repo plugins via CI.
In progress Federated learning plugin type (Flower + PyTorch); on-device training rounds; mesh-sync hooks for plugins that need to coordinate across a user's devices. Plugin signing for third-party (non-upstream) plugins.
Design Public third-party plugin index (replaces the "contribute upstream to publish" path). Plugin certification mechanism and bar. Plugin sandboxing primitives.
Research LLMMixNet integration points — how a plugin opts in to making frontier-LLM calls via the anonymous-credential mixnet rather than directly. End-to-end agentic feedback API surface for analysis plugins.

If you're building against shenas and one of these gates your work, comment on the relevant issue in shenas-org/shenas issues. Prioritization follows actual demand from people building things.