agentsview is a local-first dashboard and CLI for browsing, searching, and tracking costs across many AI coding agents, with a fast built-in replacement for ccusage.
agentsview is a single-binary tool for session intelligence and analytics around AI coding agents. It runs locally with no accounts, discovers sessions from supported agents on your machine, and serves a web UI for viewing and analyzing them. The repository also provides CLI commands for server control and usage reporting, plus options for Docker, desktop app downloads, and package installation.
The project addresses the difficulty of tracking usage, costs, and session history across many different AI coding agents in one place. It aims to replace slower tools that repeatedly re-parse raw session files by using already indexed local data, while keeping everything on-device and avoiding accounts or cloud dependency.
At a high level, agentsview scans the session directories of supported agents, synchronizes the discovered data into a local SQLite database, and then exposes that indexed information through a local server and web UI. For cost tracking, it combines session data with pricing information and supports filtering by agent, date range, and model, along with JSON output for scripting. The README also describes containerized and remote-access setups, including public URL handling and optional authentication when exposing the UI beyond loopback.
It is drawing attention because it sits at the intersection of AI coding agents, local-first tooling, and cost visibility, which are all timely needs for developers using multiple agents. The repository also highlights support for Claude Code, Codex, and more than 20 other agents, a fast local replacement for ccusage, and straightforward installation paths that lower adoption friction.
The README explicitly positions agentsview as a replacement for ccusage and similar tools. Beyond that, the clearest comparable approaches are other usage or cost trackers that work from session files, as well as agent-specific dashboards; however, the README does not name additional direct competitors. It also suggests that agentsview differs by aggregating across many agents instead of focusing on only one ecosystem.
AI-explained · grounded in each repo's README