claude-view

Mission Control for Claude Code

You have 10 AI agents running. One finished 12 minutes ago. Another hit its context limit. A third needs tool approval. You're Cmd+Tab-bing through terminals, burning $200/mo blind.

English · 繁體中文 · 简体中文 · 日本語 · Español · Français · Deutsch · Português · Italiano · 한국어 · Nederlands

curl -fsSL https://get.claudeview.ai/install.sh | sh

One command. Every session visible. Real-time.

---

What Is claude-view?

claude-view is an open-source dashboard that monitors every Claude Code session on your machine — live agents, past conversations, costs, sub-agents, hooks, tool calls — in one place. Rust backend, React frontend, ~10 MB binary. Zero config, zero accounts, local-first — your sessions, code and prompts never leave your machine (anonymous usage analytics in official builds; one-command opt-out; source builds send nothing).

50+ releases. 85 MCP tools. 9 skills. One npx claude-view.

---

Live Monitor

See every running session at a glance. No more terminal tab-switching.

Feature	What it does
Session cards	Each card shows the last message, model, cost, and status — instantly know what every agent is working on
Multi-session chat	Open sessions side-by-side in VS Code-style tabs (dockview). Drag to split horizontally or vertically
Context gauge	Real-time context window fill per session — see which agents are in the danger zone before they hit the limit
Cache countdown	Know exactly when prompt cache expires so you can time messages to save tokens
Cost tracking	Per-session and aggregate spend with token breakdown — hover for input/output/cache split by model
Sub-agent tree	See the full tree of spawned agents, their status, costs, and what tools they're calling
Notification sounds	Get pinged when a session finishes, errors, or needs your input — stop polling terminals
Multiple views	Grid, List, Kanban, Monitor, or Harness mode — pick what fits your workflow
Kanban swimlanes	Group sessions by project or branch — visual swimlane layout for multi-project workflows
Recently closed	Sessions that end appear in "Recently Closed" instead of vanishing — persists across server restarts
Queued messages	Messages waiting in the queue show as pending bubbles with a "Queued" badge
SSE-driven	All live data pushed via Server-Sent Events — eliminates stale-cache risks entirely

---

Chat & Conversation

Read, search, and interact with any session — live or historical.

Feature	What it does
Unified live chat	History and real-time messages in a single scrollable conversation — no tab-switching
Developer mode	Toggle between Chat and Developer views per session. Developer mode shows tool cards, event cards, hook metadata, and the full execution trace with filter chips
Full conversation browser	Every session, every message, fully rendered with markdown and code blocks
Tool call visualization	See file reads, edits, bash commands, MCP calls, skill invocations — not just text
Compact / verbose toggle	Skim the conversation or drill into every tool call
Thread view	Follow agent conversations with sub-agent hierarchies and indented threading
Hook events inline	Pre/post tool hooks rendered as conversation blocks — see hooks firing alongside the conversation
Export	Markdown export for context resumption or sharing
Bulk select & archive	Select multiple sessions for batch archiving with persistent filter state
Encrypted sharing	Share any session via E2E encrypted link — AES-256-GCM, zero server trust, key lives only in the URL fragment

---

Agent Internals

Claude Code does a lot behind "thinking..." that never shows in your terminal. claude-view exposes all of it.

Feature	What it does
Sub-agent conversations	Full tree of spawned agents, their prompts, outputs, and per-agent cost/token breakdown
MCP server calls	Which MCP tools are being invoked and their results
Skill / hook / plugin tracking	Which skills fired, which hooks ran, what plugins are active
Hook event recording	Dual-channel hook capture (live WebSocket + JSONL backfill) — every event recorded and browsable, even for past sessions
Session source badges	Each session shows how it was started: Terminal, VS Code, Agent SDK, or other entrypoints
Worktree branch drift	Detects when git worktree branches diverge — shown in live monitor and history
@File mention chips	@filename references extracted and shown as chips — hover for full path
Tool use timeline	Action log of every tool_use/tool_result pair with timing
Error surfacing	Errors bubble up to the session card — no buried failures
Raw message inspector	Drill into any message's raw JSON when you need the full picture

---

Search

Feature	What it does
Rust grep search	Search across all sessions — messages, tool calls, file paths. Powered by ripgrep-core over raw JSONL
Filtered search engine	SQLite pre-filter + grep search — one endpoint without maintaining a separate session search index
Project & branch filters	Scope to the project or branch you're working on right now
Command palette	Cmd+K to jump between sessions, switch views, find anything

---

Analytics

A full analytics suite for your Claude Code usage. Think Cursor's dashboard, but deeper.

Dashboard

Feature	Description
Week-over-week metrics	Session count, token usage, cost — compared to your previous period
*