[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"$f2xzr1u5prrv62":3},{"code":4,"message":5,"data":6,"timestamp":32},10000,"success",{"full_name":7,"position":8,"description":9,"language":10,"stars":11,"forks":12,"stars_gained":13,"age_days":14,"hn_threads":15,"one_liner":22,"summary_md":23,"topic":24,"github_topics":27,"license":28,"homepage":29,"updated_at":30,"history":31},"kenn-io\u002Fagentsview",13,"Local-first session search, analytics, insights, and token use statistics for coding agents, supporting Claude Code, Codex, and more than 50 other agents.","Go",6263,661,137,247,[16],{"id":17,"title":18,"points":19,"comments":20,"posted_at":21},49201062,"Agentsview – Browse, search, and track costs across all your AI coding agents",1,0,"2026-08-06","A local archive that indexes and searches every AI coding agent session on your machine, with cost tracking.","## The problem\n\nAnyone using more than one coding agent now accumulates session history in a scattering of undocumented directories, each with its own format. The transcripts are there, but nothing lets you answer basic questions: what did I ask that agent three weeks ago, which session ate the most tokens, or how much am I spending per model per project. `agentsview` addresses this by discovering sessions across supported agents, syncing them into a local SQLite database, and serving a web UI and CLI over that archive.\n\n## What's different\n\nMost people currently rely on each vendor's own history browser, which is per-agent and rarely exportable. `agentsview` instead treats transcripts as a cross-agent corpus: one index, one search box, one usage report. The two capabilities that actually set it apart are the multi-agent session discovery (the README names Claude Code, Codex, and 50+ others, plus SSH sync from remote machines without a daemon) and the cost\u002Ftoken reporting that reads from the same archive rather than from a vendor dashboard. Note that the README is unusually long and operational — install, Docker, remote access, daemon lifecycle — but thin on how robust the per-agent transcript parsers are; recent releases fixing Antigravity transcripts and repeated prompts suggest parsing is the ongoing maintenance burden.\n\n## Who it's for\n\nDevelopers who routinely switch between two or more coding agents and want a single searchable record plus spend visibility. It is also a reasonable fit for teams that want to query session data in SQL, since pushes to PostgreSQL, DuckDB and ClickHouse are supported. It is not for you if you use one agent exclusively and its built-in history suffices, nor if you need a hosted, multi-user analytics product — this is a single-user, local-first tool, and anything beyond loopback needs auth and a trusted proxy configured by hand.\n\n## Cost of adoption\n\nGo binary, MIT licensed, installable via a shell script, Homebrew cask, or Docker image; the CLI and desktop app share a local daemon. Real costs are elsewhere: the daemon runs in the background self-exiting after idle, the archive duplicates your transcripts into a SQLite file, and container deployments only see agents whose session directories you explicitly mount. Remote exposure requires `--public-url` and `--require-auth`, so misconfiguration is a plausible risk. Version numbering (v0.46.1) signals fast pre-1.0 churn.\n\n## Verdict\n\nThis is a young project: created in February 2026, roughly 247 days old, with about 6.2k stars and a recent weekly Go trending board appearance at #13. The star count is high relative to its age, which usually means a launch spike rather than settled adoption, and the Hacker News thread drew a single point, so external validation is thin. The idea is sound and the need is real for multi-agent users, but the per-agent parsing surface is large and the maintenance record is mostly bugfix releases. Try it if the problem is yours and you can tolerate churn; wait a few releases otherwise.",{"slug":25,"name":26},"developer-tools","Developer Tools",[],"MIT","https:\u002F\u002Fagentsview.io","2026-10-25T02:06:26Z",[],1792917999]