100% localZero telemetryBlocking approval gate

A solar-system command center for Claude Code agents.

Launch and govern a fleet of Claude Code agents from one tab. Solix holds every sensitive tool call until you approve it, runs each task in its own git worktree, and stops a run that blows past its budget cap — all on your machine.

Then run solix demo — a full synthetic galaxy in ~10 seconds, no setup.

See the install steps Star on GitHub

MIT licensed·Node ≥ 20·Works with any recent Claude Code

Coming next · Codex support

The same galaxy, with OpenAI Codex sessions as planets alongside Claude Code — and Maestro: describe a goal and watch your agent fleet build it, with a plan you approve first (in development).

See what ships today
The Solix galaxy — every running Claude Code agent as a planet orbiting a sun, with a pending permission flaring red.
127.0.0.1:4242 — your machine
Local onlyReal screenshotOpen source

// Views

Two views, one keystroke.

The galaxy is your at-a-glance overview — every agent, its status, and what needs you. Press V for the dense List view: a sortable table of every agent with status, cost, context %, and health, plus one-click actions. A map for the fleet; a compact board for heads-down work.

Galaxy overviewSortable listKeyboard-first
Solix galaxy view — each running Claude Code agent as a planet orbiting the sun.
Galaxy view
Screenshot coming

agent · status · cost · context % · health

List view (press V)

// Not a dashboard

Solix doesn't watch your agents. It runs them.

The galaxy is the interface — underneath it is a control plane that holds risky tool calls, isolates each task, and enforces the budget you set. All of it locally.

Blocking, not advisoryRuns on your machine

  • Blocking approval gate

    A real Claude Code hook, not a notification. The tool call is held until you answer — nothing sensitive runs behind your back and nothing is logged after the fact.

  • Isolated git worktrees

    Every task Solix launches gets its own worktree, so parallel agents never step on each other's files or on the branch you're working in.

  • Budget caps that bite

    Set a per-task cap and Solix enforces it: warn as the ring goes amber, pause the run when it hits the ceiling. Spend is a control, not a chart.

  • Two-way terminal wrapper

    Solix wraps the session instead of tailing it — so you can send input back into a live agent straight from the browser, not just watch it scroll.

  • Mirrors your claude agents

    Your existing agent definitions are read in as-is. No second config, no duplicated prompts — the crew you already have shows up in the galaxy.

// The problem

Managing AI agents is stuck in 2023.

Engineers now run 3–5 Claude Code sessions across different terminal tabs with no shared visibility — no idea which one is blocked on a permission, burning budget, or done. Solix is the missing mission control.

Before — 7 terminals, 0 visibility
After — one tab, total clarity

N tabs, 0 visibility

Five terminals open, zero idea which one is mid-build, mid-prompt, or stuck on a permission.

Context blind spots

You can't tell which agent is approaching its context limit until it derails.

Silent overspend

Opus runs quietly burn budget while you're switching windows. No alerts, no cap.

// Legend

How the metaphor works.

Every Solix concept maps to something you already see in the sky.

  • You / mission control

    → Sun

  • Claude Code session

    → Planet

  • Subagent (Task tool)

    → Moon

  • Built-in advisor

    → Inner crew-ring planet

  • Skill

    → Asteroid in the belt

  • Tool call

    → Comet streak

  • Awaiting permission

    → Red pulsing flare

  • Context usage

    → Planet size

  • Model

    → Planet color

  • Cost vs budget

    → Budget ring

  • Shared goal

    → Constellation lines

// How it works

Install → galaxy → query.

  1. Step 01

    Run solix init once

    Solix wires itself into Claude Code's hooks so it can see sessions the moment they start. No extension, no account, no cloud project.

    Full install guide
    solix — install
    $npm i -g @shmulikdav/solix
    $solix init# wires the Claude Code hooks
  2. Step 02

    Work exactly as before

    Keep launching Claude Code from whatever terminal you like — any number of tabs, any model. The galaxy, Mission Control and the advisor panel all read from the same local session data.

    The Solix galaxy view with agents as planets, subagents as moons, and skills as an asteroid belt.
  3. Step 03

    Query the fleet, don't chase tabs

    Each session becomes a planet when it starts and disappears when it ends. Ask Solix what's happening instead of reading five terminals.

    $solix status
    →3 agents running · 1 awaiting approval · $0.68 spent
    No accountNo config files

// The payoff

The action stops. Then you decide.

Solix hooks into Claude Code before the tool runs, so the whole chain is visible — which agent, which tool, what it wants — and execution stays frozen until you answer.

agent · refactor-authtool · write fileblockedyour approvalresumes
Execution is paused until you answerLocal onlyPress Y or NCost shown before you approve
Solix Mission Control panel — total spend, cost per completed mission, interventions, and agent status at a glance.
127.0.0.1:4242 — your machine, your agents, live

// Mission Control

See — and govern — what your agents cost.

Click the sun and Solix opens Mission Control: total spend, cost per completed mission, interventions, and which agents need you — the whole workspace at a glance. 100% local, so the numbers never leave your machine.

// Cost

Know what your fleet costs.

Live spend per agent and per task, a cost column in the List view, and a Mission Control overview (click the sun) with total spend and cost per completed mission. Set a budget cap and Solix stops a run that blows past it.

Solix Mission Control — total spend, cost per completed mission, interventions, and agent status.
Mission Control

// Features

Everything you need to actually run a fleet.

Local onlyMITNo account

A gate, not a notification

Solix holds every sensitive tool call and shows you the exact command or code diff before it runs — approve or deny in one keystroke. Never a blind yes/no.

Budget caps that are enforced

Every planet wears a ring of live USD spend. Set a per-task cap and Solix warns at amber — then pauses the run at the ceiling instead of letting it burn.

  • Isolated worktrees

    Each task Solix launches runs in its own git worktree, so parallel agents never collide on files or branches.

  • Two-way terminal

    Solix wraps the session, so you can send input back into a live agent from the browser — not just watch it.

  • Context fuel gauges

    Each planet shows how much context window is left, so you refactor the prompt before it derails.

  • Heartbeats

    Schedule recurring tasks every 30m / 2h / 1d as pulsing nodes.

  • Goal constellations

    Group agents under a goal; related planets link with constellation lines.

  • Timeline playback

    Scrub back through what every agent did, when, and why.

  • Mirrors your agents

    Your existing claude agent definitions load in as-is — no second config to maintain.

  • Shareable galaxies

    Export your advisor + skill setup as a JSON galaxy. One command to share.

  • Agent View bridge

    Auto-syncs with Claude Code's background sessions. Zero config.

  • Keyboard-first

    Jump between planets, approve, and mute with single keys. Never leave the keyboard.

  • Mission Control

    Click the sun for a live cost + status overview of every agent — spend, cost per mission, interventions, and what needs you.

// Meet the crew

Solix ships with a crew, not an empty canvas.

Fourteen specialist advisors — a product manager, a builder, a security auditor and more — watch your fleet and surface the things you'd miss. Five fly by default; nine more are one click away when you need them.

Enabling an advisor is free — it just joins your crew. Advisors only cost tokens when you pin one (an always-on session) or put one to work on a task — and Solix shows the spend right on each card.

opus

Compass

Product Manager

Sample finding
Caught scope creep in agent #2: 3 unrelated refactors in one PR.
On by default
opus

Forge

Builder

Sample finding
Built auth flow + tests in 4 min · 412 lines, 0 lint errors.
On by default
opus

Lumen

UX/UI Designer

Sample finding
Flagged 3 buttons below 44px tap target on /checkout.
On by default
opus

Argus

Code Reviewer

Sample finding
Blocked merge: agent #4 dropped error handling on 2 fetch calls.
On by default
opus

Sentinel

Security Auditor

Sample finding
Caught `rm -rf /` in agent #3 before approval — paused fleet.
On by default

+ 9 more advisors (opt-in, one click to enable)

  • Mira
    Test Engineer / QA
    sonnet
  • Echo
    Devrel / Docs
    sonnet
  • Helios
    Performance Engineer
    sonnet
  • Vale
    Release Engineer
    sonnet
  • Atlas
    Skill Curator
    sonnet
  • Cinder
    Debugger / Incident
    sonnet
  • Delta
    Data / DB Engineer
    sonnet
  • Spire
    Architect
    opus
  • Ledger
    FinOps / Cost
    haiku

// Why not just…

The only one that does all of this, locally.

Fair question to ask. Here's the honest matrix.

Self-reportedVerifiable in the repo
Capabilitytmux / iTermClaude /agentsCloud dashboardsSolix
Blocks the action until you approve no yes
in that one terminal
 partial yes
See every agent at once no
windows, not fleet
 no yes yes
Approve from one place no no
per-session prompts
 partial yes
Isolated worktree per task no no partial yes
Send input back to a live session yes
if you find the tab
 yes no yes
Context % per agent no partial yes yes
Cost / budget caps no
no signal
 partial yes yes
Code stays on your machine yes yes no
sent to vendor
 yes
Works with any Claude Code session yes partial no yes
Free & MIT yes yes no yes

// Quick start

Up and running in 30 seconds.

~/solix · install
$npm i -g @shmulikdav/solix
$solix install# wire Claude Code hooks
$solix start# → 127.0.0.1:4242
~/solix · live

Just want to look around first?

Run solix demo — seeds planets, comet streaks, and a pending permission.

$solix demo

Requires Node ≥ 20. Claude Code is optional — solix demo runs without it; you only need Claude Code to watch your real agents. macOS, Linux, Windows.

Full install guide View on GitHub npm package

// FAQ

Questions developers ask first.

Is the 3D just eye-candy?
No — press V for a dense, sortable table any time. The galaxy is the overview; the list is for focused work. Both are live.
Do the 14 advisors run up my API bill?
No. Enabling advisors is free — no API calls. You only spend when you pin one or invoke it on a task, and the cost is shown.
What if I run several agents on the same repo?
Launch each into its own isolated git worktree so they never overwrite each other's files.
Do I have to approve actions blind?
No — every sensitive tool call shows the exact command or diff before you approve.
Where does my code go?
Nowhere. Solix runs entirely on 127.0.0.1 — a local server + local SQLite. No account, no sign-in, no telemetry.
Read the full FAQ

// Mission control

Stop juggling terminal tabs. Start running mission control.

$npm i -g @shmulikdav/solix

Then run solix demo — a full synthetic galaxy in ~10 seconds, no setup.

Read the install guide Star on GitHub
  • MIT licensed
  • Node ≥ 20
  • 100% local
  • Zero telemetry