glm-quota-line 0.7.0 → 0.9.0

package/CHANGELOG.md

@@ -1,5 +1,23 @@
 # Changelog
 
+## 0.9.0
+
+- Replaced progressive tier-based backoff with diamond-shaped refresh strategy: high quota (80–100%) refreshes every 2 min, medium (30–79%) every 5 min, low (0–29%) every 2 min — frequent updates when usage is active or quota is near exhaustion, relaxed in between
+- Added failure-type-aware retry TTLs: rate-limited (429) retries after 3 min, unavailable retries after 2 min, instead of reusing the quota-based TTL
+- Removed `refreshCount`/`tierIndex` cache fields and `advanceTier()` — cache format is now simpler, and TTL is derived purely from quota percentage
+- Aligned severity threshold (danger/warn boundary) from hardcoded 25% to `LOW_QUOTA_THRESHOLD` (30%), consistent with the refresh band boundary
+- Unknown CLI commands and invalid config subcommands now print an error and exit with code 1 instead of silently showing quota status
+
+## 0.8.0
+
+- Added progressive refresh backoff: new sessions start at 3-minute intervals, advance to 5 minutes after 5 refreshes, and cap at 10 minutes after 5 more — giving new users fast feedback while reducing API pressure in long sessions
+- Added low-quota override: when remaining quota drops below 30%, the refresh interval is forced back to 3 minutes regardless of current tier, so users see accurate data when it matters most
+- Low-quota refreshes do not advance the tier counter, preserving the previous tier when quota recovers
+- SessionStart (startup / resume / clear) resets the tier to level 0, ensuring fresh sessions always start responsive
+- Rate-limited and failed responses no longer advance the tier counter
+- Old cache files without tier fields automatically migrate to tier 0
+- Simplified README quick start to two commands with style config clearly marked as optional
+
 ## 0.7.0
 
 - Added `--version` and `version` so the installed CLI version is visible directly from the command line

package/README.en.md

@@ -1,7 +1,7 @@
 <h1 align="center">glm-quota-line</h1>
 
 <p align="center">
-  One-command install, zero dependencies. Shows accurate GLM quota balance and reset time in the Claude Code status line — data from the official API, not estimated.
+  One-command install, zero dependencies. Shows accurate Zhipu GLM Coding Plan quota balance and reset time in the Claude Code status line — data from the official API, not estimated.
 </p>
 
 <p align="center">
@@ -15,7 +15,7 @@
   <a href="./README.md">简体中文</a>
 </p>
 
-`glm-quota-line` reads the GLM quota API, caches results, and prints a short status line for Claude Code's `statusLine.command`. A `SessionStart` hook pre-refreshes the cache so new sessions never show stale data.
+`glm-quota-line` reads the Zhipu GLM Coding Plan quota API, caches results, and prints a short status line for Claude Code's `statusLine.command`. A `SessionStart` hook pre-refreshes the cache so new sessions never show stale data.
 
 ## Features
 
@@ -25,13 +25,20 @@
 - **Zero dependencies** — no runtime deps, single-purpose CLI
 - **Smart caching** — refreshes by session, TTL, and token usage; supports manual auth overrides for proxy/gateway setups
 
-## Installation
+## Quick Start
 
 ```bash
 npm install -g glm-quota-line
+glm-quota-line install
+```
+
+That's it — works out of the box. Default output (`style=text`, `theme=ansi`, `palette=dark`):
+
+```text
+GLM Lite | 5h 91% | week 47% | reset 14:47
 ```
 
-Global install is recommended because `install` writes stable executable paths into `statusLine.command` and `SessionStart` hooks. `npx` works for one-off previews.
+Both the 5-hour quota (primary) and weekly quota (secondary) are shown with automatic reset time.
 
 Upgrade:
 
@@ -40,28 +47,30 @@ npm install -g glm-quota-line
 glm-quota-line check-update
 ```
 
-## Quick Start
+## Custom Styles
+
+All style options are optional — adjust only if needed:
 
 ```bash
-glm-quota-line install
-glm-quota-line config set style bar
-glm-quota-line config set theme ansi
-glm-quota-line config set palette dark
+glm-quota-line config set style bar       # layout: text / compact / bar
+glm-quota-line config set theme ansi      # color: plain / ansi
+glm-quota-line config set palette dark    # palette: dark / mono (only when theme=ansi)
 ```
 
-## Output
+Different `style` outputs:
 
 ```text
+# style=text (default)
 GLM Lite | 5h 91% | week 47% | reset 14:47
 
+# style=compact
 GLM 5h 91% W 47% | 14:47
 
+# style=bar
 GLM Lite █░░░░░░░░░ 91% | W 47% | 14:47
 ```
 
-Both the 5-hour quota (primary) and weekly quota (secondary) are shown with automatic reset time.
-
-## Styles & Colors
+Recommended combinations:
 
 | Use case       | Config                                        |
 | -------------- | --------------------------------------------- |
@@ -69,10 +78,6 @@ Both the 5-hour quota (primary) and weekly quota (secondary) are shown with auto
 | Light terminal | `style=compact`, `theme=ansi`, `palette=mono` |
 | Plain fallback | any `style`, `theme=plain`                    |
 
-`style` controls layout (`text` / `compact` / `bar`). `theme` enables ANSI color. `palette` only applies when `theme=ansi`.
-
-Set `NO_COLOR=1` to force plain output.
-
 ## Custom Auth
 
 If Claude Code runs behind a gateway or proxy and the injected `ANTHROPIC_AUTH_TOKEN` / `ANTHROPIC_BASE_URL` are not the real values:

package/README.md

@@ -1,7 +1,7 @@
 <h1 align="center">glm-quota-line</h1>
 
 <p align="center">
-  一行命令安装、零依赖，在 Claude Code 状态栏实时显示 GLM 配额余额与重置时间。数据来自官方接口，精准非估算。
+  一行命令安装、零依赖，在 Claude Code 状态栏实时显示智谱（Zhipu）GLM Coding Plan 配额余额与重置时间。数据来自官方接口，精准非估算。
 </p>
 
 <p align="center">
@@ -15,7 +15,7 @@
   <a href="./README.en.md">English</a>
 </p>
 
-`glm-quota-line` 读取 GLM 配额接口、缓存结果，并输出一行适合 Claude Code `statusLine.command` 的状态文本。通过 `SessionStart` hook 预刷新缓存，新会话不会显示旧配额。
+`glm-quota-line` 读取智谱（Zhipu）GLM Coding Plan 配额接口、缓存结果，并输出一行适合 Claude Code `statusLine.command` 的状态文本。通过 `SessionStart` hook 预刷新缓存，新会话不会显示旧配额。
 
 ## 功能
 
@@ -25,43 +25,52 @@
 - **零依赖** — 无运行时依赖，单一用途
 - **智能缓存** — 按会话、TTL 和 token 用量刷新；支持代理/网关手动覆盖鉴权
 
-## 安装
+## 快速开始
 
 ```bash
 npm install -g glm-quota-line
+glm-quota-line install
+```
+
+两步完成，开箱即用。默认输出（`style=text`，`theme=ansi`，`palette=dark`）：
+
+```text
+GLM Lite | 5h 91% | week 47% | reset 14:47
 ```
 
-推荐全局安装，`install` 会将稳定的可执行路径写入 `statusLine.command` 和 `SessionStart` hooks。`npx` 适合一次性试跑。
+同时显示 5 小时配额（主要）和周配额（次要），并自动展示重置时间。
 
-升级：
+升级版本：
 
 ```bash
 npm install -g glm-quota-line
 glm-quota-line check-update
 ```
 
-## 快速开始
+## 自定义样式
+
+所有样式配置均为可选，按需调整：
 
 ```bash
-glm-quota-line install
-glm-quota-line config set style bar
-glm-quota-line config set theme ansi
-glm-quota-line config set palette dark
+glm-quota-line config set style bar       # 布局: text / compact / bar
+glm-quota-line config set theme ansi      # 配色: plain / ansi
+glm-quota-line config set palette dark    # 色板: dark / mono（仅 theme=ansi 时生效）
 ```
 
-## 输出示例
+不同 style 效果：
 
 ```text
+# style=text（默认）
 GLM Lite | 5h 91% | week 47% | reset 14:47
 
+# style=compact
 GLM 5h 91% W 47% | 14:47
 
+# style=bar
 GLM Lite █░░░░░░░░░ 91% | W 47% | 14:47
 ```
 
-同时显示 5 小时配额（主要）和周配额（次要），并自动展示重置时间。
-
-## 样式与配色
+推荐搭配：
 
 | 使用场景   | 配置                                          |
 | ---------- | --------------------------------------------- |
@@ -69,10 +78,6 @@ GLM Lite █░░░░░░░░░ 91% | W 47% | 14:47
 | 浅色终端   | `style=compact`, `theme=ansi`, `palette=mono` |
 | 纯文本回退 | 任意 `style`，`theme=plain`                   |
 
-`style` 控制布局（`text` / `compact` / `bar`），`theme` 控制是否启用 ANSI 颜色，`palette` 仅在 `theme=ansi` 时生效。
-
-设置 `NO_COLOR=1` 可强制关闭颜色。
-
 ## 自定义鉴权
 
 如果 Claude Code 运行在代理或网关后面，注入的 `ANTHROPIC_AUTH_TOKEN` / `ANTHROPIC_BASE_URL` 不是实际值：
@@ -101,7 +106,7 @@ glm-quota-line config set <style|display|theme|palette|auth-token|base-url> <val
 glm-quota-line config unset <key>
 ```
 
-完整说明运行 `glm-quota-line --help`。
+运行 `glm-quota-line --help` 查看完整说明。
 
 `glm-quota-line --version` 可直接查看当前安装版本；`check-update` 只检查并提示升级命令，不会自动修改环境。
 

package/package.json

@@ -1,17 +1,19 @@
 {
   "name": "glm-quota-line",
-  "version": "0.7.0",
-  "description": "Claude Code status line helper for GLM Coding Plan quota.",
+  "version": "0.9.0",
+  "description": "Zhipu GLM Coding Plan quota monitor for Claude Code status line.",
   "type": "module",
   "license": "MIT",
   "author": "caojian",
   "preferGlobal": true,
   "keywords": [
     "claude-code",
-    "statusline",
     "glm",
+    "zhipu",
+    "z.ai",
+    "coding-plan",
     "quota",
-    "cli"
+    "statusline"
   ],
   "repository": {
     "type": "git",

package/src/claude/install.js

@@ -32,7 +32,7 @@ function isToolManagedStatusLine(statusLine) {
   return isToolManagedCommand(statusLine?.command);
 }
 
-export function getStatusLineEntryPath() {
+function getStatusLineEntryPath() {
   return path.resolve(__dirname, "..", "cli", "index.js");
 }
 
@@ -53,38 +53,41 @@ function isToolManagedHook(hook) {
   return hook?.type === "command" && isToolManagedCommand(hook.command);
 }
 
-function removeManagedSessionStartHooks(settings) {
-  if (!settings.hooks || typeof settings.hooks !== "object") {
-    return false;
-  }
-
-  const existingGroups = Array.isArray(settings.hooks.SessionStart) ? settings.hooks.SessionStart : null;
-  if (!existingGroups) {
-    return false;
-  }
+function filterManagedHooksFromGroups(groups) {
+  let hadManaged = false;
+  const filtered = [];
 
-  let removed = false;
-  const nextGroups = [];
-
-  for (const group of existingGroups) {
+  for (const group of groups) {
     if (!group || typeof group !== "object" || !Array.isArray(group.hooks)) {
-      nextGroups.push(group);
+      filtered.push(group);
       continue;
     }
 
     const nextHooks = group.hooks.filter((hook) => !isToolManagedHook(hook));
     if (nextHooks.length !== group.hooks.length) {
-      removed = true;
+      hadManaged = true;
     }
 
     if (nextHooks.length > 0) {
-      nextGroups.push({
-        ...group,
-        hooks: nextHooks
-      });
+      filtered.push({ ...group, hooks: nextHooks });
     }
   }
 
+  return { groups: filtered, hadManaged };
+}
+
+function removeManagedSessionStartHooks(settings) {
+  if (!settings.hooks || typeof settings.hooks !== "object") {
+    return false;
+  }
+
+  const existingGroups = Array.isArray(settings.hooks.SessionStart) ? settings.hooks.SessionStart : null;
+  if (!existingGroups) {
+    return false;
+  }
+
+  const { groups: nextGroups, hadManaged: removed } = filterManagedHooksFromGroups(existingGroups);
+
   if (nextGroups.length > 0) {
     settings.hooks.SessionStart = nextGroups;
   } else {
@@ -101,21 +104,10 @@ function removeManagedSessionStartHooks(settings) {
 function installManagedSessionStartHooks(settings, command) {
   const existingGroups =
     settings.hooks && typeof settings.hooks === "object" && Array.isArray(settings.hooks.SessionStart)
-      ? settings.hooks.SessionStart.map((group) =>
-          group && typeof group === "object" ? { ...group, hooks: Array.isArray(group.hooks) ? [...group.hooks] : group.hooks } : group
-        )
+      ? settings.hooks.SessionStart
       : [];
 
-  const nextGroups = existingGroups.map((group) => {
-    if (!group || typeof group !== "object" || !Array.isArray(group.hooks)) {
-      return group;
-    }
-
-    return {
-      ...group,
-      hooks: group.hooks.filter((hook) => !isToolManagedHook(hook))
-    };
-  });
+  const { groups: nextGroups } = filterManagedHooksFromGroups(existingGroups);
 
   for (const matcher of SESSION_START_MATCHERS) {
     const targetGroup = nextGroups.find(
@@ -123,21 +115,13 @@ function installManagedSessionStartHooks(settings, command) {
     );
 
     if (targetGroup) {
-      targetGroup.hooks.push({
-        type: "command",
-        command
-      });
+      targetGroup.hooks.push({ type: "command", command });
       continue;
     }
 
     nextGroups.push({
       matcher,
-      hooks: [
-        {
-          type: "command",
-          command
-        }
-      ]
+      hooks: [{ type: "command", command }]
     });
   }
 

package/src/claude/sessionStart.js

@@ -3,7 +3,7 @@ import { resolveQuotaStatus } from "../core/quota/service.js";
 import { readStatusLineInput } from "./input.js";
 import { readToolConfig } from "./settings.js";
 
-export const SESSION_START_MATCHERS = ["startup", "resume", "clear", "compact"];
+export const SESSION_START_MATCHERS = ["startup", "resume", "clear"];
 
 export async function refreshQuotaOnSessionStart(options = {}) {
   const hookInput = await readStatusLineInput(options.stdin ?? process.stdin);
@@ -12,10 +12,7 @@ export async function refreshQuotaOnSessionStart(options = {}) {
   const resolveQuotaStatusFn = options.resolveQuotaStatusFn ?? resolveQuotaStatus;
   const config = {
     ...loadConfigFn(options.env ?? process.env, userConfig),
-    sessionId: hookInput?.session_id || "",
-    // SessionStart runs before statusLine has token totals, so treat a new session
-    // as starting from zero to preserve later token-threshold refreshes.
-    observedTokens: 0
+    sessionId: hookInput?.session_id || ""
   };
 
   return resolveQuotaStatusFn(config, {

package/src/claude/settings.js

@@ -10,8 +10,9 @@ import {
   isValidTheme
 } from "../shared/constants.js";
 import { readJsonFile, writeJsonFile } from "../shared/jsonFile.js";
+import { normalizeOptionalString } from "../shared/config.js";
 
-export function getClaudeDir() {
+function getClaudeDir() {
   return path.join(os.homedir(), ".claude");
 }
 
@@ -23,15 +24,6 @@ export function getToolConfigPath() {
   return path.join(getClaudeDir(), "glm-quota-line.json");
 }
 
-function normalizeOptionalString(value) {
-  if (typeof value !== "string") {
-    return undefined;
-  }
-
-  const trimmed = value.trim();
-  return trimmed || undefined;
-}
-
 function redactSecret(value) {
   if (!value) {
     return value;

package/src/cli/commands.js

@@ -149,11 +149,13 @@ export async function handleCommand(args, output = process.stdout, dependencies
   if (command === "config" && subcommand === "set") {
     const configKey = CONFIG_KEYS[key];
     if (!configKey) {
+      process.exitCode = 1;
       output.write("Supported config keys: style, display, theme, palette, auth-token, base-url\n");
       return true;
     }
 
     if (!configKey.validate(value)) {
+      process.exitCode = 1;
       output.write(`${configKey.invalidMessage}\n`);
       return true;
     }
@@ -168,6 +170,7 @@ export async function handleCommand(args, output = process.stdout, dependencies
   if (command === "config" && subcommand === "unset") {
     const configKey = CONFIG_KEYS[key];
     if (!configKey) {
+      process.exitCode = 1;
       output.write("Supported config keys: style, display, theme, palette, auth-token, base-url\n");
       return true;
     }
@@ -177,5 +180,17 @@ export async function handleCommand(args, output = process.stdout, dependencies
     return true;
   }
 
+  if (command === "config") {
+    process.exitCode = 1;
+    output.write("Supported config subcommands: show, set, unset\n");
+    return true;
+  }
+
+  if (command) {
+    process.exitCode = 1;
+    output.write(`Unknown command: ${command}\nRun 'glm-quota-line -h' for usage.\n`);
+    return true;
+  }
+
   return false;
 }

package/src/cli/index.js

@@ -67,14 +67,6 @@ Examples:
 Environment:
   ANTHROPIC_AUTH_TOKEN
   ANTHROPIC_BASE_URL
-  GLM_DISPLAY_MODE
-  GLM_STYLE
-  GLM_BAR_WIDTH
-  GLM_TIMEOUT_MS
-  GLM_CACHE_TTL_MS
-  GLM_THEME
-  GLM_PALETTE
-  NO_COLOR
 `);
 }
 
@@ -87,21 +79,6 @@ function getStoredDisplayOverrides(userConfig) {
   };
 }
 
-function asNonNegativeNumber(value) {
-  return Number.isFinite(value) && value >= 0 ? value : null;
-}
-
-function getObservedTokens(statusLineInput) {
-  const inputTokens = asNonNegativeNumber(statusLineInput?.context_window?.total_input_tokens);
-  const outputTokens = asNonNegativeNumber(statusLineInput?.context_window?.total_output_tokens);
-
-  if (inputTokens === null && outputTokens === null) {
-    return null;
-  }
-
-  return (inputTokens ?? 0) + (outputTokens ?? 0);
-}
-
 export async function main() {
   try {
     const args = parseArgs();
@@ -128,8 +105,7 @@ export async function main() {
       // Display config precedence is env defaults -> persisted config -> CLI flags.
       ...getStoredDisplayOverrides(userConfig),
       ...args,
-      sessionId: statusLineInput?.session_id || "",
-      observedTokens: getObservedTokens(statusLineInput)
+      sessionId: statusLineInput?.session_id || ""
     };
     const quotaStatus = await resolveQuotaStatus(config);
 

package/src/cli/update.js

@@ -1,6 +1,6 @@
+import { DEFAULT_TIMEOUT_MS } from "../shared/constants.js";
 import { readPackageInfo } from "../shared/packageInfo.js";
 
-const DEFAULT_TIMEOUT_MS = 5000;
 const DEFAULT_REGISTRY_URL = "https://registry.npmjs.org";
 
 function parseVersion(version) {

package/src/core/quota/cache.js

@@ -1,6 +1,8 @@
 import fs from "node:fs/promises";
 import path from "node:path";
 
+import { asFiniteNumber } from "./parse.js";
+
 function isValidQuotaShape(value) {
   if (!value || typeof value.key !== "string") {
     return false;
@@ -69,10 +71,6 @@ function isSuccessCacheShape(value) {
   return isMultiQuotaSuccessCacheShape(value) || isLegacyPercentSuccessCacheShape(value);
 }
 
-function asFiniteNumber(value) {
-  return Number.isFinite(value) ? value : null;
-}
-
 function normalizeCache(parsed) {
   if (!parsed || typeof parsed !== "object") {
     return null;
@@ -81,9 +79,11 @@ function normalizeCache(parsed) {
   const result = isSuccessCacheShape(parsed.result) ? parsed.result : null;
   const savedAt = asFiniteNumber(parsed.savedAt);
   const lastAttemptAt = asFiniteNumber(parsed.lastAttemptAt) ?? savedAt;
-  const lastObservedTokensAtFetch = asFiniteNumber(parsed.lastObservedTokensAtFetch);
   const sessionId = typeof parsed.sessionId === "string" ? parsed.sessionId : "";
-  const skipNextTokenTrigger = parsed.skipNextTokenTrigger === true;
+  const lastFailureKind =
+    parsed.lastFailureKind === "rate_limited" || parsed.lastFailureKind === "unavailable"
+      ? parsed.lastFailureKind
+      : null;
 
   if (result && savedAt === null) {
     return null;
@@ -96,9 +96,8 @@ function normalizeCache(parsed) {
   return {
     savedAt,
     lastAttemptAt,
-    lastObservedTokensAtFetch,
     sessionId,
-    skipNextTokenTrigger,
+    lastFailureKind,
     result
   };
 }
@@ -109,10 +108,7 @@ async function writeCache(cacheFilePath, cache) {
       ...(Number.isFinite(cache.savedAt) ? { savedAt: cache.savedAt } : {}),
       ...(Number.isFinite(cache.lastAttemptAt) ? { lastAttemptAt: cache.lastAttemptAt } : {}),
       ...(cache.sessionId ? { sessionId: cache.sessionId } : {}),
-      ...(Number.isFinite(cache.lastObservedTokensAtFetch)
-        ? { lastObservedTokensAtFetch: cache.lastObservedTokensAtFetch }
-        : {}),
-      ...(cache.skipNextTokenTrigger ? { skipNextTokenTrigger: true } : {}),
+      ...(cache.lastFailureKind ? { lastFailureKind: cache.lastFailureKind } : {}),
       ...(cache.result ? { result: cache.result } : {})
     },
     null,
@@ -139,34 +135,19 @@ export async function writeSuccessCache(cacheFilePath, result, options = {}) {
     savedAt: now,
     lastAttemptAt: now,
     sessionId: options.sessionId || "",
-    lastObservedTokensAtFetch: asFiniteNumber(options.observedTokens),
-    skipNextTokenTrigger: false,
+    lastFailureKind: null,
     result
   });
 }
 
-export async function writeRateLimitedCache(cacheFilePath, cached, options = {}) {
+export async function writeFailureCache(cacheFilePath, cached, options = {}) {
   const now = options.now ?? Date.now();
 
   await writeCache(cacheFilePath, {
     savedAt: cached?.savedAt ?? null,
     lastAttemptAt: now,
     sessionId: options.sessionId || cached?.sessionId || "",
-    lastObservedTokensAtFetch:
-      asFiniteNumber(options.observedTokens) ?? cached?.lastObservedTokensAtFetch ?? null,
-    skipNextTokenTrigger: true,
-    result: cached?.result ?? null
-  });
-}
-
-export async function writeSkippedTokenTriggerCache(cacheFilePath, cached, options = {}) {
-  await writeCache(cacheFilePath, {
-    savedAt: cached?.savedAt ?? null,
-    lastAttemptAt: cached?.lastAttemptAt ?? null,
-    sessionId: options.sessionId || cached?.sessionId || "",
-    lastObservedTokensAtFetch:
-      asFiniteNumber(options.observedTokens) ?? cached?.lastObservedTokensAtFetch ?? null,
-    skipNextTokenTrigger: false,
+    lastFailureKind: options.failureKind || "unavailable",
     result: cached?.result ?? null
   });
 }

package/src/core/quota/parse.js

@@ -1,4 +1,4 @@
-function asFiniteNumber(value) {
+export function asFiniteNumber(value) {
   return Number.isFinite(value) ? value : null;
 }
 

package/src/core/quota/service.js

@@ -1,74 +1,74 @@
 import {
   readCache,
-  writeRateLimitedCache,
-  writeSkippedTokenTriggerCache,
+  writeFailureCache,
   writeSuccessCache
 } from "./cache.js";
 import { fetchQuota } from "./fetch.js";
 import { parseQuotaResponse } from "./parse.js";
-
-const TOKEN_REFRESH_THRESHOLD = 300_000;
-
-function normalizeObservedTokens(value) {
-  return Number.isFinite(value) && value >= 0 ? value : null;
-}
+import {
+  RATE_LIMIT_RETRY_TTL_MS,
+  REFRESH_BANDS,
+  UNAVAILABLE_RETRY_TTL_MS
+} from "../../shared/constants.js";
 
 function getCachedStatus(cached) {
   return cached?.result ?? { kind: "unavailable" };
 }
 
-function shouldRefreshQuota(cached, cacheTtlMs, now, sessionId, observedTokens) {
-  if (!cached) {
-    return { kind: "fetch" };
+function getRefreshTtlMs(leftPercent) {
+  if (!Number.isFinite(leftPercent)) {
+    return REFRESH_BANDS[1].ttlMs;
   }
 
-  if (sessionId && cached.sessionId !== sessionId) {
-    return { kind: "fetch" };
+  const matchedBand = REFRESH_BANDS.find((band) => leftPercent >= band.minLeftPercent);
+  return matchedBand ? matchedBand.ttlMs : REFRESH_BANDS[REFRESH_BANDS.length - 1].ttlMs;
+}
+
+function getEffectiveTtlMs(cached) {
+  if (cached?.lastFailureKind === "rate_limited") {
+    return RATE_LIMIT_RETRY_TTL_MS;
   }
 
-  if (cached.lastAttemptAt === null || now - cached.lastAttemptAt >= cacheTtlMs) {
-    return { kind: "fetch" };
+  if (cached?.lastFailureKind === "unavailable") {
+    return UNAVAILABLE_RETRY_TTL_MS;
   }
 
-  if (observedTokens === null || cached.lastObservedTokensAtFetch === null) {
-    return { kind: "cache" };
+  return getRefreshTtlMs(cached?.result?.leftPercent);
+}
+
+function shouldRefreshQuota(cached, cacheTtlMs, now, sessionId) {
+  if (!cached) {
+    return true;
   }
 
-  if (observedTokens - cached.lastObservedTokensAtFetch < TOKEN_REFRESH_THRESHOLD) {
-    return { kind: "cache" };
+  if (sessionId && cached.sessionId !== sessionId) {
+    return true;
   }
 
-  if (cached.skipNextTokenTrigger) {
-    return { kind: "skip_token_trigger" };
+  if (cached.lastAttemptAt === null || now - cached.lastAttemptAt >= cacheTtlMs) {
+    return true;
   }
 
-  return { kind: "fetch" };
+  return false;
 }
 
 export async function resolveQuotaStatus(config, options = {}) {
   const now = options.now ?? Date.now();
   const fetchImpl = options.fetchImpl;
   const sessionId = config.sessionId || "";
-  const observedTokens = normalizeObservedTokens(config.observedTokens);
 
   if (!config.authorization) {
     return { kind: "auth_error" };
   }
 
   const cached = await readCache(config.cacheFilePath);
-  const refreshDecision = options.forceRefresh
-    ? { kind: "fetch" }
-    : shouldRefreshQuota(cached, config.cacheTtlMs, now, sessionId, observedTokens);
+  const effectiveTtlMs = getEffectiveTtlMs(cached);
 
-  if (refreshDecision.kind === "cache") {
-    return getCachedStatus(cached);
-  }
+  const shouldRefresh = options.forceRefresh
+    ? true
+    : shouldRefreshQuota(cached, effectiveTtlMs, now, sessionId);
 
-  if (refreshDecision.kind === "skip_token_trigger") {
-    await writeSkippedTokenTriggerCache(config.cacheFilePath, cached, {
-      sessionId,
-      observedTokens
-    });
+  if (!shouldRefresh) {
     return getCachedStatus(cached);
   }
 
@@ -78,8 +78,7 @@ export async function resolveQuotaStatus(config, options = {}) {
   if (parsed.kind === "success") {
     await writeSuccessCache(config.cacheFilePath, parsed, {
       now,
-      sessionId,
-      observedTokens
+      sessionId
     });
     return parsed;
   }
@@ -89,17 +88,27 @@ export async function resolveQuotaStatus(config, options = {}) {
   }
 
   if (parsed.kind === "rate_limited") {
-    await writeRateLimitedCache(config.cacheFilePath, cached, {
+    await writeFailureCache(config.cacheFilePath, cached, {
       now,
       sessionId,
-      observedTokens
+      failureKind: "rate_limited"
     });
     return getCachedStatus(cached);
   }
 
   if (cached?.result) {
+    await writeFailureCache(config.cacheFilePath, cached, {
+      now,
+      sessionId,
+      failureKind: "unavailable"
+    });
     return cached.result;
   }
 
+  await writeFailureCache(config.cacheFilePath, cached, {
+    now,
+    sessionId,
+    failureKind: "unavailable"
+  });
   return parsed;
 }

package/src/core/status/bar.js

@@ -1,6 +1,6 @@
 import { DEFAULT_BAR_WIDTH, STATUS_BAR_CHARACTERS } from "../../shared/constants.js";
 
-export function normalizeBarWidth(barWidth) {
+function normalizeBarWidth(barWidth) {
   if (!Number.isFinite(barWidth)) {
     return DEFAULT_BAR_WIDTH;
   }

package/src/core/status/format.js

@@ -8,10 +8,6 @@ import { buildBar } from "./bar.js";
 import { applyTheme } from "./theme.js";
 import { buildStatusViewModel } from "./viewModel.js";
 
-function getSeverityTone(model) {
-  return model.severity || "neutral";
-}
-
 function createErrorSegments(model) {
   const tone = model.kind === "auth_error" ? "danger" : "warn";
   return [
@@ -75,7 +71,7 @@ function appendResetSegments(segments, model) {
 }
 
 function createTextSegments(model, displayMode) {
-  const severityTone = getSeverityTone(model);
+  const severityTone = model.severity;
 
   return appendResetSegments(
     appendSecondarySegments(
@@ -91,7 +87,7 @@ function createTextSegments(model, displayMode) {
 }
 
 function createCompactSegments(model) {
-  const severityTone = getSeverityTone(model);
+  const severityTone = model.severity;
   let segments;
 
   if (model.secondaryQuota) {
@@ -126,7 +122,7 @@ function createBarSegments(model, barWidth) {
   }
 
   const bar = buildBar(model.primaryQuota.usedPercent, barWidth);
-  const severityTone = getSeverityTone(model);
+  const severityTone = model.severity;
   const segments = [
     { text: model.levelLabel, tone: "label" },
     { text: " ", tone: "plain" },
@@ -157,8 +153,7 @@ export function formatStatus(result, options = {}) {
   if (model.kind !== "success") {
     return applyTheme(createErrorSegments(model), {
       theme: normalizeTheme(options.theme),
-      palette: normalizePalette(options.palette),
-      env: options.env
+      palette: normalizePalette(options.palette)
     });
   }
 
@@ -175,7 +170,6 @@ export function formatStatus(result, options = {}) {
 
   return applyTheme(segments, {
     theme: normalizeTheme(options.theme),
-    palette: normalizePalette(options.palette),
-    env: options.env
+    palette: normalizePalette(options.palette)
   });
 }

package/src/core/status/theme.js

@@ -17,15 +17,6 @@ const ANSI = {
   red: "\u001b[31m"
 };
 
-function shouldUseAnsi(theme, env = process.env) {
-  // NO_COLOR is the explicit opt-out and wins over local config.
-  if (env?.NO_COLOR) {
-    return false;
-  }
-
-  return normalizeTheme(theme || DEFAULT_THEME) === "ansi";
-}
-
 function applyCodes(text, codes) {
   if (!text || !codes.length) {
     return text;
@@ -76,9 +67,8 @@ function getMonoCodes(tone) {
 export function applyTheme(segments, options = {}) {
   const theme = normalizeTheme(options.theme || DEFAULT_THEME);
   const palette = normalizePalette(options.palette || DEFAULT_PALETTE);
-  const env = options.env || process.env;
 
-  if (!shouldUseAnsi(theme, env)) {
+  if (theme !== "ansi") {
     return segments.map((segment) => segment.text).join("");
   }
 

package/src/core/status/viewModel.js

@@ -1,3 +1,5 @@
+import { LOW_QUOTA_THRESHOLD } from "../../shared/constants.js";
+
 function padTwoDigits(value) {
   return String(value).padStart(2, "0");
 }
@@ -37,7 +39,7 @@ function getSeverity(leftPercent) {
     return "good";
   }
 
-  if (leftPercent >= 25) {
+  if (leftPercent >= LOW_QUOTA_THRESHOLD) {
     return "warn";
   }
 

package/src/shared/config.js

@@ -12,22 +12,9 @@ import {
   DEFAULT_QUOTA_URL,
   DEFAULT_STYLE,
   DEFAULT_THEME,
-  DEFAULT_TIMEOUT_MS,
-  normalizeDisplayMode,
-  normalizePalette,
-  normalizeStatusStyle,
-  normalizeTheme
+  DEFAULT_TIMEOUT_MS
 } from "./constants.js";
 
-function parsePositiveInt(value, fallback) {
-  if (value === undefined || value === null || value === "") {
-    return fallback;
-  }
-
-  const parsed = Number.parseInt(String(value), 10);
-  return Number.isFinite(parsed) && parsed > 0 ? parsed : fallback;
-}
-
 function getCacheRoot() {
   if (process.platform === "darwin") {
     return path.join(os.homedir(), "Library", "Caches");
@@ -68,7 +55,7 @@ function deriveQuotaUrl(baseUrl) {
   return "";
 }
 
-function normalizeOptionalString(value) {
+export function normalizeOptionalString(value) {
   if (typeof value !== "string") {
     return "";
   }
@@ -92,13 +79,13 @@ export function loadConfig(env = process.env, overrides = {}) {
     quotaUrl: derivedQuotaUrl || DEFAULT_QUOTA_URL,
     authorization,
     anthropicBaseUrl,
-    timeoutMs: parsePositiveInt(env.GLM_TIMEOUT_MS, DEFAULT_TIMEOUT_MS),
-    cacheTtlMs: parsePositiveInt(env.GLM_CACHE_TTL_MS, DEFAULT_CACHE_TTL_MS),
-    displayMode: normalizeDisplayMode(env.GLM_DISPLAY_MODE || DEFAULT_DISPLAY_MODE),
-    style: normalizeStatusStyle(env.GLM_STYLE || DEFAULT_STYLE),
-    theme: normalizeTheme(env.GLM_THEME || DEFAULT_THEME),
-    palette: normalizePalette(env.GLM_PALETTE || DEFAULT_PALETTE),
-    barWidth: parsePositiveInt(env.GLM_BAR_WIDTH, DEFAULT_BAR_WIDTH),
+    timeoutMs: DEFAULT_TIMEOUT_MS,
+    cacheTtlMs: DEFAULT_CACHE_TTL_MS,
+    displayMode: DEFAULT_DISPLAY_MODE,
+    style: DEFAULT_STYLE,
+    theme: DEFAULT_THEME,
+    palette: DEFAULT_PALETTE,
+    barWidth: DEFAULT_BAR_WIDTH,
     cacheFilePath: path.join(getCacheRoot(), "glm-quota-line", cacheFileName)
   };
 }

package/src/shared/constants.js

@@ -2,18 +2,23 @@ export const DEFAULT_QUOTA_URL = "https://bigmodel.cn/api/monitor/usage/quota/li
 export const DEFAULT_CN_BASE_URL = "https://open.bigmodel.cn";
 export const DEFAULT_INTL_BASE_URL = "https://api.z.ai";
 export const DEFAULT_TIMEOUT_MS = 5000;
-export const DEFAULT_CACHE_TTL_MS = 300_000;
+export const DEFAULT_CACHE_TTL_MS = 600_000;
+
+export const REFRESH_BANDS = [
+  { minLeftPercent: 80, ttlMs: 120_000 }, // 2 min
+  { minLeftPercent: 30, ttlMs: 300_000 }, // 5 min
+  { minLeftPercent: 0, ttlMs: 120_000 } // 2 min
+];
+
+export const LOW_QUOTA_THRESHOLD = 30;
+export const RATE_LIMIT_RETRY_TTL_MS = 180_000;
+export const UNAVAILABLE_RETRY_TTL_MS = 120_000;
 export const DEFAULT_DISPLAY_MODE = "left";
 export const DEFAULT_STYLE = "text";
 export const DEFAULT_BAR_WIDTH = 10;
-export const DEFAULT_THEME = "plain";
+export const DEFAULT_THEME = "ansi";
 export const DEFAULT_PALETTE = "dark";
 
-export const STATUS_STYLES = ["text", "compact", "bar"];
-export const DISPLAY_MODES = ["left", "used", "both"];
-export const THEMES = ["plain", "ansi"];
-export const PALETTES = ["dark", "mono"];
-
 export const STATUS_BAR_CHARACTERS = {
   filled: "█",
   empty: "░"
@@ -22,23 +27,20 @@ export const STATUS_BAR_CHARACTERS = {
 export const TOOL_CONFIG_SCHEMA_VERSION = 1;
 export const TOOL_CONFIG_MANAGED_BY = "glm-quota-line";
 
-export const QUOTA_UNAVAILABLE_TEXT = "GLM | quota unavailable";
-export const AUTH_EXPIRED_TEXT = "GLM | auth expired";
-
 export function isValidStatusStyle(value) {
-  return STATUS_STYLES.includes(value);
+  return value === "text" || value === "compact" || value === "bar";
 }
 
 export function isValidDisplayMode(value) {
-  return DISPLAY_MODES.includes(value);
+  return value === "left" || value === "used" || value === "both";
 }
 
 export function isValidTheme(value) {
-  return THEMES.includes(value);
+  return value === "plain" || value === "ansi";
 }
 
 export function isValidPalette(value) {
-  return PALETTES.includes(value);
+  return value === "dark" || value === "mono";
 }
 
 export function normalizeStatusStyle(value) {