Jupiter v0.1 Alpha: First Public Release

What shipped

After months of development, we’re releasing Jupiter v0.1 Alpha — the first public version of our parallel AI coding agent orchestrator.

Here’s what’s in this release:

Planner engine

The Planner is Jupiter’s brain. Before any worker touches your code, the Planner:

Indexes your codebase — symbols, modules, dependencies
Decomposes tasks — breaks a high-level instruction into worker-sized subtasks
Generates briefings — each worker gets a context package with relevant code, constraints, and owned symbols
Detects conflicts — ensures no two workers modify the same code

Parallel workers

Jupiter workers are AI coding agents enhanced with MCP tools. In v0.1:

Up to 4 parallel workers per task
Each worker gets a typed briefing from the Planner
Workers share state via the memory bus
Automatic conflict detection and resolution

MCP ecosystem

16 Rust-native MCP servers integrated:

Phase 1: Filesystem, code + git + system tools, Cargo toolchain
Phase 2: Code intelligence (narsil-mcp), Rust LSP, Rust docs
Phase 3: Web fetch, SQLite, SSH
Phase 4: Sequential thinking, error tracking, crate search

All servers are single binaries, zero Node.js dependencies, ~3MB RAM each.

What’s next

Here’s our roadmap for the coming releases:

v0.2 — Auto-validation: workers run tests and self-correct before reporting done
v0.3 — CLI + GUI: unified interface for managing workers and tasks
v0.4 — Remote workers: distribute across machines via SSH + SSE transport
v1.0 — Production-ready with full documentation

Getting started

Jupiter is open source under AGPL-3.0. Join our Discord to connect with the community and follow the development.

Thank you

Jupiter is built by a small team that believes AI coding agents can be dramatically better. If you share that vision, we’d love your feedback, bug reports, and contributions.

Let’s build the future of parallel AI development — together.