AgentsView¶

A local-first desktop and web app for browsing, searching, and analyzing your past AI coding sessions. See where your agents' time and money actually go — across every project, model, and tool.

Get Started View on GitHub

Reads sessions from dozens of AI coding agents — auto-discovered, nothing to configure.

Quick Start¶

Download the desktop app (recommended):

Download the latest .dmg (macOS), .exe (Windows), or .AppImage (Linux) from GitHub Releases or via homebrew: brew install --cask agentsview. The desktop app is fully bundled and includes built-in auto-update.

Install via pip — or run instantly with uvx:

pip install agentsview    # install permanently
uvx agentsview            # or run without installing

Install via shell script:

curl -fsSL https://agentsview.io/install.sh | bash

Windows (PowerShell):

powershell -ExecutionPolicy ByPass -c "irm https://agentsview.io/install.ps1 | iex"

agentsview serve              # Start server
agentsview serve --port 9090  # Custom port
agentsview serve --no-browser # Disable browser auto-open
agentsview serve --background # Run in the background

Note

The desktop app and CLI share the same data directory (~/.agentsview/), so you can use one or both — they are fully complementary.

Fast Token Usage & Cost Reports¶

If you've been reaching for ccusage to see how much you spent on Claude Code yesterday, try agentsview usage instead. It reads from the same pre-indexed SQLite database that powers the rest of AgentsView, so reports come back in well under a second even on large histories. It reports on token-bearing sessions from Claude Code, Codex, Copilot CLI, OpenCode-format tools, Pi, Gemini, Qwen Code, OpenClaw/QClaw, Hermes, WorkBuddy, Forge, Piebald, Antigravity, Zed, VS Code Copilot, Visual Studio Copilot, gptme, Mistral Vibe, and more as parser coverage expands.

agentsview usage daily          # last 30 days, terminal table
agentsview usage daily --all    # full history, JSON-friendly
agentsview usage statusline     # $9.61 today

On a 22,000-session local database, agentsview usage daily runs 80–220× faster than npx ccusage@latest daily (see benchmarks). On smaller databases the absolute gap is smaller, but reports still come back sub-second. See Token Usage & Costs for the full write-up.

See When Your Agents Are Working¶

The Activity dashboard turns timestamped session data into a clear picture of when your agents ran, how much work overlapped, and what it cost. See peak concurrency and the exact moment it happened, active versus idle time, agent-minutes across concurrent sessions, and total cost — scoped to any day, week, month, or custom range and filterable by project, agent, and machine.

AgentsView Activity dashboard

Click any bucket in the concurrency timeline to see exactly which sessions were running in that slot, overlay token or cost trends over the bars, and break activity down by project, model, or agent. The same report is available from the CLI, with --json for scripting:

agentsview activity report --preset day
agentsview activity report --preset week --json

See Activity for the full reference.

What It Does¶

AgentsView reads the session files that your AI coding agents leave on your machine and gives you a local-first desktop and web app to work with them. By default everything stays on your machine. Optionally, PostgreSQL sync can push session data to a shared database for team or multi-machine setups.

AI-Powered Insights

Generate summaries and analysis of your coding sessions using Claude, Codex, Copilot, or Gemini. Get daily activity digests, multi-day analyses, and recommendations — scoped by project or across everything.
Browse Sessions

Scroll through every session across all your projects. See the full conversation: user prompts, assistant responses, thinking blocks, and tool calls. Filter by project, agent, date, or message count.
Search Everything

Full-text search across all message content. Find that one conversation where you discussed a specific function, error message, or design decision — even months later.
Analyze Your Usage

Activity heatmaps, tool usage breakdowns, velocity metrics, session-health analytics, per-project stats, and session distribution charts. Understand how you use agents over time.
Activity & Concurrency

See when your agents were actually working, how much ran in parallel, and what it cost — peak concurrency, active versus idle time, agent-minutes, and cost over any time window, on the Activity page.
Token Usage & Costs

A sub-second agentsview usage CLI for daily spend reports and a today's-cost status line. A ccusage alternative for token-bearing sessions across multiple agents — including Claude Code, Codex, Copilot CLI, VS Code Copilot, and Zed — that runs 80–220× faster on large session histories.
Live Sync

Watches your session directories for changes and streams new messages in real time. Start a coding session in one window, watch it appear in AgentsView in another.
Multi-Agent Support

Works with dozens of AI coding session sources including Claude Code, Codex, Copilot, Cursor, Gemini, OpenHands, Aider, Claude Cowork, DeepSeek TUI, gptme, Kilo, MiMoCode, Mistral Vibe, OhMyPi, QwenPaw, Reasonix, Shelley, and Visual Studio Copilot. Auto-discovers session directories so there's nothing to configure.
Import Chat History

Import your Claude.ai and ChatGPT conversations — including images. Upload a zip export and browse everything in one place alongside your agent coding sessions.
Runs Locally

SQLite database, embedded web frontend, no cloud services, no accounts. Install the desktop app or a single binary and run it.

How It Works¶

AgentsView architecture: agent sessions sync into SQLite with FTS5 search, served via REST API, SSE events, and embedded Svelte SPA

AgentsView watches your agent session directories for changes, parses JSONL files from each agent format, and stores structured data in SQLite with full-text search indexes. The embedded web frontend provides browsing, search, and analytics over the REST API.