---

What it does

Claude Code is a remarkable CLI. But the moment you have more than one project, more than one account, or more than one machine, the experience fragments. You lose track of which terminal is which, which sessions are paused, what they have spent, and where their attention is going.

Claude Command Center (CCC) wraps Claude Code and Codex in a desktop app that treats the session as the first-class object. Every session has a colour, a name, an account, a working directory, and a saved config. Every spawn surfaces its tokens, its model, its rate-limit window, and its identity. Every cent is captured and pivotable.

It does not replace Claude Code. It conducts it.

---

Highlights

Sessions and saved configs

Every workspace starts as a saved config: a label, a colour, a working directory, a model, and any agent templates you want pre-loaded. New sessions land on Opus 4.8. Effort is a live setting you change inside Claude with /effort and read off the statusline and the session card, not a config field. Drag a folder onto the sidebar to bootstrap a working-directory config in one drop. Local or SSH, the same form drives both.

Multiple accounts, per-session isolation

Run more than one Claude account side by side. Your existing login is captured into a protected primary account on first run, and every session runs under a saved, isolated account, so signing in to one never disturbs another or your default login. You pick the account at launch time, the first time a session spawns this run. Add an account by running /login in any session: CCC detects the new login and offers to save it as a separate named account. Name and colour each one in Settings, Accounts; the colour follows the account onto the session card, the statusline, and the launch picker. Memory, settings, and history stay shared across all accounts.

Logs, a chat-transcript viewer

CCC indexes Claude's own conversation transcripts and renders them back as a readable chat: messages, tool calls, and thinking. A timeline rail beside the transcript scrubs the whole conversation; click to jump. Full-text search spans every conversation and jumps you straight to the matching turn. A per-session Conversation tab live-follows the running session. Deleting the index never touches your conversations, which stay in ~/.claude/projects.

Tokenomics

Track every dollar Claude and Codex cost you across every session. A background indexer reads all of your transcripts (including subagent and sidechain files), dedups globally, and computes cost at query time from live pricing, so the dashboard opens instantly with a KPI row (total spend, tokens, sessions, daily burn), a daily-spend chart and per-model breakdown, and a sessions table with cost, model, and config attribution. Filter the whole view by date, model, account, or project. Pricing comes from LiteLLM open pricing (cached 24 hours).

Memory dashboard

A dashboard over Claude's auto-memory across every project. A KPI strip (memories, projects, total size, stale-over-30-days, index health) and charts summarise the whole store; a ranked project list shows staleness dots, index warnings, and live-session activity. Drill into any project for a sortable memory table, then open a memory in the reading drawer to read it cleanly, write missing frontmatter, or delete it. Full-text search runs across everything. Catches drift before it bloats your context.

---

The rest of the surface

CCC Sentinel An opt-in watcher that notices when Claude Code updates and checks whether the new version might affect CCC. Findings surface in a labelled Sentinel chip in the title bar and a panel. It proposes model and effort registry fixes you apply yourself and never changes anything automatically. A hot-reloadable registry means brand-new models still get a colour, label, and pricing. Fail-open, so it never blocks the app. Toggle it in Settings, CCC Sentinel.	Conductor MCP A local MCP server exposing 17 browser-vision tools (screenshot, navigate, click, type, eval) plus codex_review, so Claude can ask Codex to spot-check its own work. One global Chrome instance is shared across every session, so cookies and logins persist. SSH sessions reach the same server transparently through an auto-injected reverse tunnel.
Agent Hub Two surfaces in one. Tasks dispatch headless Claude as background jobs with live status and output streaming. The Library is where you author agent templates (name, prompt, model, tool whitelist) that surface as tickable subagents in every Edit Config dialog, so Claude can delegate to them via the Task tool inside a running session. Teams chain agents with shared context and per-step prompts.	Codex provider, first-class OpenAI Codex CLI sits alongside Claude in the New Session dialog. Pick the provider per spawn. The gpt-5 series sits in the model dropdown, with read-only / standard / auto / unrestricted permission presets in the toolbar. The resume picker mirrors the Claude flow. Tokenomics segments Codex spend automatically.
GitHub PR context where you work A collapsible right-rail panel surfaces PR status, CI runs, reviews, unresolved threads, and inferred issue context from your branch and transcript. Sign in via OAuth or PAT, or adopt your existing gh CLI auth. Per-session opt-in; toggles with Ctrl+/ (Cmd+/ on macOS).	Combined Mode and Draw Pair Claude with a partner shell (pwsh, bash, cmd) in the same tab for the build, git, and docker commands you want one keystroke from your prompt. The Draw scratchpad is a per-session Excalidraw whiteboard that persists with the config; freeze the webview pane to annotate a snapshot, or export the canvas straight into Claude as an image.
Snap and Vision capture Region or window capture from any screen, encoded at 1920px / JPEG 85 to stay under Claude's image budget. Local sessions get a file path inline; SSH fetches the image over the Conductor MCP tunnel. Paste images from the clipboard with Alt+V.	Dynamic workflows, surfaced Invoke an Opus 4.8 dynamic workflow three ways: include the word workflow in a prompt for a one-off, run /effort ultracode so every substantive task auto-orchestrates, or run the bundled /deep-research <question>. Watch active runs with /workflows. CCC rolls workflow spend into your session totals, and a Settings toggle disables dynamic workflows globally if you want them off.

---

Install

Download

1. Grab the latest installer from Releases.
   • Windows: ClaudeCommandCenter-x.y.z.exe
   • macOS (Apple Silicon): ClaudeCommandCenter-x.y.z-mac.dmg
2. Verify the SHA-256 of the file you downloaded against the checksum printed on the release page.
3. Run the installer and choose your Data and Resources directories.
4. The setup wizard hands off to Claude CLI auth.

Requirements

• Claude Code CLI installed and authenticated (a recent version, for Opus 4.8 and dynamic workflows)
• Node.js 20+ (a Claude Code dependency)
• Windows 10 or 11 (x64), or macOS 12+ on Apple Silicon (arm64)

Windows SmartScreen · These installers are not code-signed. This is an independent, single-maintainer build, and a code-signing certificate is not in place yet, so Windows SmartScreen will warn the first time you run a new release: click More info, then Run anyway. To convince yourself the download is intact, verify the SHA-256 against the value on the release page before running it. Releases produced by the CI pipeline are additionally scanned through VirusTotal (70+ engines); the scan link is included in those release notes.

macOS · The DMG is not notarised. Gatekeeper will block it on first open: right-click the app and choose Open, or allow it under System Settings, Privacy & Security.

---

Build from source

git clone https://github.com/nubbymong/claude-command-center.git
cd claude-command-center
npm install
npm run typecheck    # tsc, no emit
npx vitest run       # unit suite
npm run dev          # HMR dev
npm run build        # production build

npm run package:win  # Windows NSIS installer
npm run package:mac  # macOS DMG (Apple Silicon)

The repository ships well over two thousand unit tests plus a native (better-sqlite3 / node-pty) suite; both run green on Windows and macOS in CI on every labelled PR. See the Actions tab for current status.

---

Under the hood

Layer	Stack
Shell	Electron (frameless, sandboxed renderer, zod-validated IPC)
UI	React 18 + Tailwind CSS v4 (@theme tokens, Catppuccin Mocha by default, with light and system themes)
State	Zustand 5 (hydrated from disk on boot)
Terminal	xterm.js 5.5 + node-pty (ConPTY on Windows)
Build	electron-vite
MCP	@modelcontextprotocol/sdk (Conductor MCP server: browser vision + Codex review)
Tests	Vitest unit + native suites, Playwright E2E

The main process owns config persistence, the PTY pool, the hooks HTTP gateway (which drives the session attention pulse), the tokenomics aggregator, the statusline ingest, the Conductor MCP server, and cloud-agent dispatch. The renderer is a React SPA that hydrates from disk and talks to main exclusively through typed IPC channels. SSH sessions get a per-session settings file, a per-session MCP config, and a reverse tunnel injected automatically.

---

Security

Layer	What we do
Credentials	SSH passwords, sudo passwords, and encrypted notes are stored as encrypted blobs via the OS keystore (DPAPI on Windows, Keychain on macOS, libsecret on Linux). Machine-bound, never plaintext.
Account isolation	Each session runs under its own isolated home so signing in to one account never touches another or your default login. The original global login is snapshotted read-only on first run.
Permissions	CCC honors Claude Code's own permission prompts and settings. It is not a gate and never auto-approves or intercepts tool calls on your behalf.
Telemetry	None of our own. The Claude API goes through the Claude CLI directly. Outbound is limited to: GitHub Releases (update check), status.claude.com (status pills), the GitHub API (opt-in GitHub sidebar after sign-in), and LiteLLM open-pricing JSON (cached 24 hours).
Data integrity	Atomic config writes (.tmp + rename). Daily snapshots of CONFIG/*.json to CONFIG/_backups/YYYY-MM-DD/, 7-day retention. Sandboxed renderer; typed IPC with schema validation on data-bearing channels.
Releases	Installers are unsigned (see Install). Releases built by the CI pipeline are scanned through VirusTotal across 70+ engines; verify the SHA-256 from the release page before running any installer.

Report vulnerabilities privately via GitHub Security Advisories. See SECURITY.md for scope.

---

Keyboard shortcuts

Shortcut	Action
Ctrl+T	New config
Ctrl+W	Close session
Ctrl+Tab / Ctrl+Shift+Tab	Next / previous session
Ctrl+1 … Ctrl+9	Jump to session N
Ctrl+B	Toggle sidebar
Ctrl+/ (Cmd+/ on macOS)	Toggle GitHub sidebar
Alt+V	Paste clipboard image
Escape	Interrupt Claude
Shift+Enter	New line without sending

All shortcuts are rebindable in Settings, Shortcuts.

---

Project history

CCC was developed privately from late 2025 and open-sourced in April 2026. The git history was squashed for the initial public release; everything from v1.0 forward is in the open.

Per-release detail lives in src/renderer/changelog.ts.

---

Contributing

Contributions welcome. See CONTRIBUTING.md for setup, coding standards, and the PR process.

---

Disclaimer and trademarks

Claude and Claude Code are trademarks of Anthropic, PBC. OpenAI and Codex are trademarks of OpenAI. This project is an independent, community-built tool. It is not affiliated with, endorsed by, sponsored by, or supported by Anthropic or OpenAI.

All references to "Claude", "Claude Code", "Codex", "Anthropic", or "OpenAI" are nominative, used solely to identify the third-party software this tool interoperates with. Claude Command Center is a separate work that wraps and orchestrates the official Claude Code and Codex CLIs. It does not include, modify, or redistribute their code, and it requires you to install and authenticate those tools yourself under their own terms.

If you are a rights holder with a concern about this project's use of a name or mark, please open a GitHub issue or contact the maintainer and it will be addressed promptly.

---

License

MIT.