Feature: non-interactive way to read plan quota (5-hour / weekly / monthly)
まだ誰も着手していません。
評価
調査の方向性
Start by tracing the CLI's existing projectUsageView and the documented-in-the-issue endpoints: /alpha/whoami?limits=1, /alpha/billing/credits, /alpha/billing/subscriptions, and /alpha/usage/summary?since=. Define the supported command output around the five-hour, weekly, and monthly fields, then verify that the text and --json forms exit non-interactively and expose the requested values.
索引モデルが issue の本文から書いたものです。
説明
Feature Description
Make the plan's quota readable from the CLI in a non-interactive way — one command that prints the 5-hour, weekly and monthly windows and exits, with a --json form for scripts.
Concretely, something like:
$ command-code usage
GOAT / alex
5h ▌░░░░░░░░░ 5% $13.32 left of $14.00 resets 04:16 (in 3h35m)
week ███▊░░░░░░ 37% $22.00 left of $35.00 resets 09-27 23:29 (in 5d22h)
month █▉░░░░░░░░ 19% $57.00 left of $70.00 resets 10-20 23:25 (in 28d22h)
and command-code usage --json printing the same fields (used, cap, percent, remaining, resetAt, resetsInMs) for tooling.
Right now the numbers exist — the dashboard shows them, and the web app reads them from /alpha/billing/credits (credits + windowLimits.fiveHour/weekly) and /alpha/billing/subscriptions — but as far as I can tell there is no supported way to ask for them without a browser. If a non-interactive path already exists, a pointer to it in the docs would be just as good as a new subcommand.
Optional, if it fits your roadmap: the same numbers in the terminal title or prompt, so they are ambient rather than something you remember to ask for. That is what most people build on top of this themselves.
Use Case
The quota is what decides whether work can continue, and it is checked at exactly the moments the user is inside a terminal:
- Sizing a task before starting it. "Can I finish this refactor on the current 5-hour window, or do I spread it over two?" needs the window's remaining amount and its reset countdown, not a percentage on a page in another tab.
- Explaining a failure. When a request is refused mid-run, the first question is whether the window is spent and when it comes back. Today that answer means leaving the editor for a dashboard.
- Scripts and status lines. CI jobs, hooks and status-line integrations can only consume the quota if there is a documented, non-interactive output contract. Today they have to depend on
/alpha/*— the web app's own endpoints — which are undocumented and can change without notice. A supported--jsonoutput would turn that from a hack into an interface. - Team plans. The same read with an
orgIdanswers "how much does the team have left", which is currently only visible per-account in the app.
Reading the quota should also cost nothing: it is four GETs, no model call.
Additional Context
I hit this while building a usage line for the agents I work in (Claude Code, Codex, ZCode, opencode, pi, DeepSeek Harness, and Grok) and ended up implementing it as an external plugin because there was no first-party surface to read:
- https://github.com/Jovan1666/commandcode-usage — MIT, ~300 lines of Node with no dependencies, and it is exactly the read described above: resolve the credential (
COMMAND_CODE_API_KEYor~/.commandcode/auth.json), call/alpha/whoami?limits=1,/alpha/billing/credits,/alpha/billing/subscriptions,/alpha/usage/summary?since=, normalize, print. - The normalization is written to match the CLI's own
projectUsageView: the plan's face value is the denominator while the subscription isactive, otherwise it falls back to spent + remaining;windowLimits.limited: falsemeans the account has no rolling windows and the plain monthly pool is the whole story. - A few details I had to guess at, which a first-party implementation could settle: whether
windowLimits.*.capis ever absent for a plan (I keep a plan table as a fallback), whatexceededimplies for subsequent requests, and whetheraverageCostis the right basis for a request count estimate (it moves with the model mix, so any count is an estimate).
Happy to hand over the code, or to open a PR against this repo if you want it there rather than in a plugin.
How important is this to you?
Important for my workflow
- 主要言語
- 言語のデータがありません
- スター
- 4k
- フォーク
- 350
- PR マージ指標
- 30日以内にマージされた PR はありません
コントリビューションガイド
このリポジトリのコントリビューションガイドは索引されていません
はじめの一歩
- issue を最後まで読み、次にプロジェクトのコントリビューションガイドを読みます。
- 着手することを issue にコメントします — 二人が同じ作業をするのを防げます。
- リポジトリをフォークし、ブランチを切って変更します。
- issue 番号を参照したプルリクエストを送ります。
CommandCodeAI/command-code のほかの issue
-
難易度 2/5 1〜3時間 初心者へのやさしさ 68/100
CommandCodeAI/command-code#903 · コメント 2 件 ·
-
難易度 2/5 1〜3時間 初心者へのやさしさ 68/100
CommandCodeAI/command-code#855 ·
-
難易度 2/5 1〜3時間 初心者へのやさしさ 78/100
CommandCodeAI/command-code#841 · コメント 1 件 ·
-
難易度 2/5 1〜3時間 初心者へのやさしさ 68/100
CommandCodeAI/command-code#655 · コメント 1 件 ·
-
難易度 2/5 1〜3時間 初心者へのやさしさ 68/100
CommandCodeAI/command-code#608 ·
CommandCodeAI/command-code の issue をすべて見る
似ている issue
-
難易度 2/5 1〜3時間 初心者へのやさしさ 78/100
syfoud/Simulated_Scepter#172 ·
-
難易度 1/5 1時間未満 初心者へのやさしさ 82/100
-
難易度 2/5 1〜3時間 初心者へのやさしさ 88/100
crossplane/crossplane#7859 ·
-
bot:ai-assisted component:compact-js status:untriaged
難易度 2/5 1〜3時間 初心者へのやさしさ 84/100
midnightntwrk/midnight-sdk#403 ·
-
難易度 2/5 1〜3時間 初心者へのやさしさ 78/100
nightscout/nocturne#1379 ·