agentsview

agentsview

A local-first desktop and web application for browsing, searching, and analyzing AI agent coding sessions. Supports Claude Code, Codex, OpenCode, and 8 other agents.

Desktop App

Download the desktop installer for macOS or Windows from GitHub Releases. The desktop app includes auto-updates and runs the server as a local sidecar -- no terminal required.

CLI Install

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

Windows:

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

The CLI installer downloads the latest release, verifies the SHA-256 checksum, and installs the binary.

Build from source (requires Go 1.25+ with CGO and Node.js 22+):

git clone https://github.com/wesm/agentsview.git
cd agentsview
make build
make install  # installs to ~/.local/bin

Why?

AI coding agents generate large volumes of session data across projects. agentsview indexes these sessions into a local SQLite database with full-text search, providing a web interface to find past conversations, review agent behavior, and track usage patterns over time.

Features

• Full-text search across all message content, instantly
• Analytics dashboard with activity heatmaps, tool usage, velocity metrics, and project breakdowns
• Multi-agent support for Claude Code, Codex, OpenCode, and 8 other agents (full list)
• Live updates via SSE as active sessions receive new messages
• Keyboard-first navigation (vim-style j/k/[/])
• Export and publish sessions as HTML or to GitHub Gist
• Local-first -- all data stays on your machine, single binary, no accounts

Usage

agentsview              # start server
agentsview -port 9090   # custom port

On startup, agentsview discovers sessions from all supported agents, syncs them into a local SQLite database with FTS5 full-text search, and opens a web UI at http://127.0.0.1:8080.

For hostname or reverse-proxy access, set a public_url. This preserves the default DNS-rebinding and CSRF protections while explicitly trusting the external browser origin you expect.

# Direct HTTP on a custom hostname/port
agentsview -host 0.0.0.0 -port 8004 \
  -public-url http://viewer.example.test:8004

# HTTPS behind your own reverse proxy
agentsview -host 127.0.0.1 -port 8004 \
  -public-url https://viewer.example.test

agentsview can also manage a Caddy frontend for you. In managed-Caddy mode, keep the backend on loopback and let Caddy terminate TLS and optionally restrict client IP ranges. By default, managed Caddy binds to 127.0.0.1 and exposes the public URL on port 8443. To expose it on a non-loopback interface, set -proxy-bind-host explicitly and provide at least one -allowed-subnet.

Managed Caddy mode requires the caddy CLI to already be installed. This patch does not automate Caddy installation. Use your normal OS package manager or ask your coding agent to install Caddy for your platform first. Caddy supports Linux, macOS, and Windows.

For privileged ports such as 443 or 80, prefer leaving agentsview itself unprivileged and granting the Caddy binary permission to bind low ports. On Linux, that typically means:

sudo setcap cap_net_bind_service=+ep "$(command -v caddy)"

Then run agentsview normally as your user with -public-port 443 or -public-port 80. This avoids running the session viewer as root, which would otherwise change which home directory and agent session data it can see. If you do not need a privileged port, the default 8443 is the simpler option.

agentsview -host 127.0.0.1 -port 8080 \
  -public-url https://viewer.example.test \
  -proxy caddy \
  -proxy-bind-host 0.0.0.0 \
  -public-port 8443 \
  -tls-cert ~/.certs/viewer.crt \
  -tls-key ~/.certs/viewer.key \
  -allowed-subnet 10.0/16 \
  -allowed-subnet 192.168.1.0/24

You can persist the same settings in ~/.agentsview/config.json:

{
  "public_url": "https://viewer.example.test",
  "proxy": {
    "mode": "caddy",
    "bind_host": "0.0.0.0",
    "public_port": 8443,
    "tls_cert": "/home/user/.certs/viewer.crt",
    "tls_key": "/home/user/.certs/viewer.key",
    "allowed_subnets": [
      "10.0/16",
      "192.168.1.0/24"
    ]
  }
}

public_origins remains available as an advanced override when you need to allow additional browser origins beyond the main public_url.

Screenshots

Dashboard	Session viewer

Search	Activity heatmap

Keyboard Shortcuts

Documentation

Full documentation is available at agentsview.io:

• Quick Start -- installation and first run
• Usage Guide -- dashboard, session browser, search, export
• CLI Reference -- commands, flags, and environment variables
• Configuration -- data directory, config file, session discovery
• Architecture -- how the sync engine, parsers, and server work

Development

make dev            # run Go server in dev mode
make frontend-dev   # run Vite dev server (use alongside make dev)
make desktop-dev    # run Tauri desktop app in dev mode
make test           # Go tests (CGO_ENABLED=1 -tags fts5)
make lint           # golangci-lint
make e2e            # Playwright E2E tests

Desktop Development

The desktop app is a Tauri wrapper under desktop/. It launches the agentsview Go binary as a local sidecar and loads http://127.0.0.1:<port> in a native webview.

make desktop-dev                 # run desktop app in dev mode
make desktop-build               # build desktop bundles (.app/.exe)
make desktop-macos-app           # build macOS .app only
make desktop-windows-installer   # build Windows installer (.exe)

Desktop env escape hatch: ~/.agentsview/desktop.env (for PATH/API keys overrides).

Project Structure

cmd/agentsview/     CLI entrypoint
internal/config/    Configuration loading
internal/db/        SQLite operations (sessions, search, analytics)
internal/parser/    Session parsers (all supported agents)
internal/server/    HTTP handlers, SSE, middleware
internal/sync/      Sync engine, file watcher, discovery
frontend/           Svelte 5 SPA (Vite, TypeScript)

Supported Agents

Acknowledgements

Inspired by claude-history-tool by Andy Fischer and claude-code-transcripts by Simon Willison.

License

MIT