What we're building
We ship frequently and build in public. Here's where Orbit stands and where it's headed.
Claude SDK, sub-agents, configurable interaction (plan-first or direct execution)
CodeMirror 6 with full LSP support, syntax highlighting, autocomplete
Agent takes screenshots, navigates autonomously, click-to-select any element
Tabbed interface with full agent access for commands and scripts
Markdown editor for docs and PRDs — agent reads as context while building
Staging, commits, branches, diffs, history, push/pull
Create custom skills, plugins, and MCP servers. 20K+ community skills via skills.sh
Sign in with any Claude plan — Free, Pro, or Max. No API key required
Bring your own Anthropic API key, no token markup
Quick actions for common agent tasks
OpenAI, Google, and local models alongside Claude
Native builds for all major platforms
Tab-autocomplete as you type
Tester, PR Reviewer (4-pass), Research, DevOps, CI
Curated sets for frontend, backend, databases, system design
Design surface for spatial thinking
Ship to production directly from Orbit
Ollama, LMStudio — fully offline, fully private
Structured build → review → deploy cycles
Not in V1 — by design
V1 is agent-first: full context, every surface, one conversation. These are coming — we'd rather ship them right than ship them now.
Shape what we build next
We prioritize based on what users actually need. Tell us what matters to you.