Behind every "thinking..." spinner, Claude is spawning sub-agents, calling MCP servers, running skills, firing hooks — and you can't see any of it.
You're paying $200/mo for Claude Code. You deserve a dashboard.
curl -fsSL https://get.claudeview.ai/install.sh | sh
One command. Every session visible. Real-time.
Install
| Shell (recommended) | curl -fsSL https://get.claudeview.ai/install.sh | sh |
| npx | npx claude-view |
What You Get
Live Monitor
- Live session cards — see what every session is working on, right now
- Notification sounds — get pinged when a session finishes or needs input
- Context gauge — real-time context window usage per session
- Cache warm countdown — time your messages to save tokens
- Cost tracking — per-session and aggregate spend with cache savings
- Sub-agent visualization — see the full agent tree, tool calls, MCP invocations
- Recently closed sessions — sessions stay visible after ending instead of vanishing
- Unified live chat — history and real-time messages in one scrollable view
- SSE-driven live data — real-time push, no stale cache
History & Search
- Rich chat history — every conversation rendered with markdown, code blocks, tool calls
- Rust grep search — ripgrep-core + SQLite pre-filter across sessions, messages, tool calls, file paths
- Export — markdown export for context resumption or sharing
- Encrypted sharing — share any session via E2E encrypted link
Plans, Prompts & Teams
- Plans browser — view your
.claude/plans/ directly in session detail
- Prompt history — full-text search across all prompts with intent classification and template clustering
- Teams dashboard — track team leads, inbox, tasks, and file changes across team members
Plugin Manager
- GUI plugin browser — install, enable, disable, and uninstall Claude Code plugins — no terminal needed
- Marketplace dialog — discover and install plugins with user or project scope
Workflows
- Workflow builder — create and run multi-stage workflows with a Mermaid diagram preview and YAML editor
- Streaming LLM chat rail — generate workflow definitions in real time
Open in IDE
- One-click file open — open any referenced file directly in VS Code, Cursor, Zed, or your preferred editor
- Auto-detects your editor — no configuration needed
Agent Internals
- @File mention chips —
@filename references shown as chips on session cards
- Agent SDK live chat — thinking blocks, tool calls, and results with syntax highlighting
- Worktree branch drift — detects when git worktree branches diverge
Analytics
- Activity heatmap — 90-day GitHub-style usage grid
- Cost ROI metrics — cost per commit, per session, per line of AI output
- AI Fluency Score — single 0–100 number tracking your overall effectiveness
How It Works
On first run, npx claude-view downloads a platform-specific Rust binary (~10 MB) from GitHub Releases. The binary is cached at ~/.claude-view/ so subsequent runs start instantly.
Your sessions, code, and prompts stay on your machine — claude-view reads and indexes them entirely locally. To guide development, official builds (this npm package and the install script) send anonymous, content-free usage analytics by default (which features are used, app version, OS — never code, prompts, paths, or session content). A one-time notice prints on first run; opt out anytime with CLAUDE_VIEW_TELEMETRY=0. Builds from source send nothing. See the Privacy Policy.
Configuration
CLAUDE_VIEW_PORT | 47892 | Port for the local server |
PORT | 47892 | Alternative port override |
Supported Platforms
| macOS | Apple Silicon (arm64), Intel (x64) |
| Linux | x64 |
| Windows | x64 |
Links