Show HN: A Claude usage menu bar small enough to read before you run it
A Show HN post titled “A Claude usage menu bar small enough to read before you run it” points to tokease, a project by tpatrouillat on GitHub at…
By Dillip Chowdary • Aug 07, 2026 • Source: HN Claude/Codex/Fable
A Show HN post titled “A Claude usage menu bar small enough to read before you run it” points to tokease, a project by tpatrouillat on GitHub at github.com/tpatrouillat/tokease. The thread is at news.ycombinator.com/item?id=49210250 and, at the time captured here, had 2 points and 1 comment. The pitch is narrow: a menu bar surface for Claude usage that stays small enough to inspect before you run anything.
Technically, the product is framed as a usage monitor, not a full client or IDE plugin. It sits in the menu bar and surfaces Claude usage so you can read the state before launching work that would burn tokens. The repo name tokease and the “small enough to read before you run it” line both stress a compact surface area: check usage first, then act. No architecture diagram, benchmark, or release number was supplied with the submission data, so the mechanics stop at that menu bar plus pre-run visibility claim.
Advertisement
Tech Pulse Daily
Get tomorrow's pulse first
Join engineers who read Tech Pulse before stand-up. Free, weekday mornings.
For engineers and builders already paying for Claude (and often running Codex-style agent loops in the same stack), a always-visible usage read reduces the usual failure mode of discovering spend only after a long session. Pre-run checks matter when prompts, tools, and multi-step agents make cost nonlinear. A menu bar tool fits the “glance and go” habit better than opening a billing page or parsing logs mid-flow.
On the market side, this sits in the same HN cluster as Claude, Codex, and Fable-adjacent tooling: lightweight utilities around model spend and local workflow, not another chat UI. Early traction on the thread is thin (2 points, 1 comment), so this is a discovery-stage Show HN rather than a crowded category leader. Competing approaches are usually dashboards, CLI quota checks, or provider console pages; a dedicated menu bar app is a different UX bet—always present, low ceremony.
Practical takeaway: if you run Claude from the desktop and want usage visible before a costly run, clone or try tokease from the GitHub repo and judge whether the menu bar read is enough for your workflow. Watch whether the single HN comment thread grows into feature requests (providers, multi-account, hard limits) and whether the author ships a clear install path and usage definition on the repo. Until more detail lands in the README or thread, treat the product as a small pre-run usage glance, not a full billing system.
Advertisement
🔎 More interesting news
- Announcing Cloudflare Ambassadors, Community Engineers, and another $1M in open-source…
- Claude Code context management: when to /clear and when to /compact
- I gave a Claude Fable 5 agent a domain and $90 it can't spend without me
- In-terminal browser inside a local sandbox for Claude Code
- Today's full Tech Pulse briefing →