Skip to main content

Roadmap

User-facing, version-sliced. Internal design rationale lives in docs/spec/11-version-roadmap.md.

Status legend:

  • Shipped — available in the current release.
  • In progress — landed in main, not yet in a tagged release, or actively being built.
  • Planned — committed to a specific milestone.
  • Considered — on the list, not committed.

0.7 (current) — feature-complete​

Shipped:

  • Workspace discovery with zero-config auto-detection.
  • Content-hash bump planner with Blake3 prerequisites and TTLs.
  • Release engine: plan / approve / apply / rollback for every strategy.
  • Policy engine: declarative TOML + sandboxed Luau, waivers with mandatory expiry.
  • MCP server: rmcp, stdio + local HTTP, ~10 tools.
  • BYO-API-key AI overlay: Anthropic / OpenAI / Gemini / Ollama.
  • Versiond daemon: JSON-RPC 2.0, file-watch cache invalidation.
  • TUI dashboard.
  • Cross-repo fleet orchestration with saga protocol.
  • 30 crates, 280+ tests.

0.8 — hardening​

Planned:

  • Windows parity audit across every subsystem. Close every "Unix-only" corner.
  • Cross-platform CI matrix expansion (Windows ARM64, Linux musl on ARM64).
  • Error-message pass. Every user-facing error ends with an actionable suggestion.
  • Doctor command (versionx doctor) covers every known failure mode.
  • Performance pass on versionx status for large monorepos (500+ packages).

Considered:

  • Per-package task caching (scaffolding for the local cache landing in 1.2).

0.9 — ecosystem breadth​

Planned:

  • Go adapter (Tier 2 → Stable).
  • Ruby adapter (Tier 2 → Stable).
  • Go runtime installer.
  • Ruby runtime installer (rv primary, ruby-build fallback).

Considered:

  • Bundler (Ruby) migration from Gemfile.lock.

1.0 — stable baseline​

Planned:

  • Every subsystem tagged Stable (no Experimental badges on Reference pages).
  • Reusable GitHub Actions published to a stable org.
  • Cross-repo saga tested end-to-end by an outside user on ≥20 repos.
  • docs.rs API reference complete for the whole SDK surface.
  • API and config stability guarantees documented.

Target: late 2026. No committed date.

1.1​

Planned:

  • OCI adapter (container-image-based artifacts).
  • JVM runtime installer (Temurin default, foojay Disco API).
  • Versioned docs on this site (snapshot 1.0 and 1.1).

1.2​

Planned:

  • JVM adapter (Maven / Gradle).
  • Local task cache in versionx-tasks.
  • Remote state backend (Postgres).
  • Optional bearer-auth for the HTTP API.

Considered:

  • Hosted GitHub App (SaaS). Still a business question.

2.0​

Planned:

  • Remote task cache.
  • Large-fleet optimizations (multi-user shared daemon, systemd/launchd units).

Considered:

  • Extension SDK for third-party adapters / policies / release strategies as first-class out-of-tree plugins.

Explicitly deferred​

  • Telemetry of any kind. Never.
  • Bundled LLM. Never. BYO API key or run your own Ollama.
  • Replacing a resolver. We drive real package managers; we don't reimplement them.

How this page evolves​

The roadmap updates on release PRs. Badges move from Planned to In progress to Shipped as work lands. If you care about a specific item and want to know what's blocking it, open a discussion.

See also​