Plan. Assign.
Watch agents deliver.
Built for agents. Clear for humans.
Tasks, ideas, issues, docs — organized into Orbits. Agents pick up work via MCP or CLI. You get a clean dashboard.
you: Pick up the next task
Starting task #42: "Implement JWT auth with refresh tokens"
Instructions loaded from spec auth-architecture
Branch: feature/task-42-auth · Status: in_progress
you: I found a bug — login fails on mobile Safari
Issue #18 registered: "Login fails on mobile Safari" (severity: high)
Linked to task #42. Dashboard updated.
you: Done. What's next?
Task #42 marked done. Next up: Task #43 "Add rate limiting to auth endpoints"
Priority: medium · Dependencies: clear_
The six pillars
Organize your AI agent workflow.
Tasks
Agents pick up, execute, and complete work with subtasks, dependencies, and instructions.
Issues
Lightweight bug tracking. Discovered during work, registered immediately, resolved later.
Ideas
Capture a thought in 5 seconds, evaluate later, convert the best ones into tasks.
Vibes
Ad-hoc fixes, exploratory coding, technical discoveries — your project's unplanned activity stream.
Docs
Architecture specs and design decisions accessible to agents across projects.
Orbits
Group tasks, issues, and docs across multiple repos into one coherent work package.
Just tell your agent
Natural language. Real actions.
See how an agent connects to Orbitmap, understands your context, and executes your workflow.
❯ What's next on my plate?
Info Reading Orbitmap context...
Found top task: Setup billing
Loaded docs: Stripe API spec
❯ I noticed a bug in auth. Working on it.
Logged issue #12: Auth glitch
Task marked In Progress
Just now
New Issue #12 Created
11 sec ago
Vibe Logged
1 min ago
How it works
From zero to agent-managed project in 5 minutes
Three steps. No complex configuration.
Start a new project
Create tasks, set priorities, and attach specs in the dashboard.
Connect your IDE
Add one JSON block (MCP) or install the CLI. 30 seconds.
Say "pick up my next task"
Your agent connects, reads the spec, and starts working.
Option A: MCP Config (recommended)
{
"mcpServers": {
"orbitmap": {
"url": "https://mcp.orbitmap.ai",
"headers": {
"Authorization": "Bearer YOUR_KEY"
}
}
}
}
Works with Claude Code, Cursor, Windsurf, VS Code + Copilot
Option B: CLI (any terminal agent)
$ npx orbitmap init --key YOUR_KEY
# Works with any agent:
# Codex CLI, Gemini CLI, Aider,
# or any terminal-based tool
For agents and IDEs that don't support MCP natively
Integrations
Works with your tools
Orbitmap uses the open MCP standard. No vendor lock-in.
Who is it for
Built for how you actually work.
Whether you manage a fleet of autonomous agents or you're a solo dev moving at the speed of thought, Orbitmap adapts to your flow.
Teams building with AI agents
Give your agents structured context — tasks, docs, and instructions — so they work autonomously instead of asking you what to do next. Ensure your whole team has visibility into what every agent is touching.
Vibe coders
Ship fast without losing track. Capture ideas, log discoveries, and keep your exploratory coding organized — all without ever leaving your terminal or breaking your flow state.
PMs & Owners
See what agents actually did — not just what was assigned. Real-time visibility into tasks, decisions, and progress.
FAQ
Common questions
I already use Linear / Jira. Do I need to switch?
Is it only for Claude / Anthropic?
npx orbitmap)
works with literally anything.
What if I'm a solo developer?
My agent can already manage tasks via chat. Why do I need this?
How is this different from GitHub Issues + Projects?
Is my code / data safe?
Your agents forget.
Orbitmap remembers.
Give your agents memory, structure, and a place to work. Start free — no credit card, no trial period.