context-monitor
Skill2 starsChanged 4 months agoMITView on GitHub
在 Claude Code 状态栏实时显示 context 使用百分比, 逼近上限时变色预警, 让你在"中途爆 context"之前主动 /compact。一键安装: 用 jq 增量写 settings.json (不覆盖既有配置, 改前自动备份), 并正确处理 /compact 边界(postTokens)、 跳过 API 出错留下的全 0 合成消息、按 1M/200k 自动判断上下文上限。 触发场景: 监控 context / 上下文用量、状态栏显示 token、防止上下文超限、 "context 满了来不及 compact"、配置 status line。即使用户没明说"状态栏", 只要想"看见"上下文用量、或防止会话中途爆掉, 都应使用本 skill。 Installs a Claude Code status-line meter that shows live context-window usage with color warnings, so you can /compact before overflowing mid-session.
2 other projects use a version of this.
Adopted by
Most recently taken up by
- context-dot-dev/cursor-plugintheir own version, first seen Oct 15, 2026
- coyvalyss1/model-matchmakertheir own version, first seen Mar 12, 2026
Other versions2 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 2
- coyvalyss1/model-matchmaker★ 168same name
- context-dot-dev/cursor-plugin★ 9same name
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository11 other files it gives its agents.
- agnes-aiSkill
- auto-approveSkill
- blog-pipelineSkill
- curator-pipelineSkill
- grok-imageSkill
- handoffSkill
- opportunity-radarSkill
- public-apisSkill
- readme-writerSkill
- sentinelSkill
- smart-fetchSkill
The file
Files
context-monitor — Claude Code 上下文用量状态栏
在状态栏实时显示 context 占用,绿/黄/红三档预警,避免会话中途突然爆 context 来不及 /compact。
Opus 4.8 ██░░░░░░░░ 18% (180k/1000k)为什么需要它
Claude Code 有 autocompact 兜底,但你看不见 context 逼近的过程——往往是"突然报超限"才发现,这时已经来不及主动 /compact。这个 skill 给状态栏加一个实时计量条,让你在变黄、变红时就主动下手。
安装
前置:python3、jq。
bash ~/.claude/skills/context-monitor/scripts/install.sh做的事:把脚本复制到 ~/.claude/context-monitor/、用 jq 增量写入 settings.json 的 statusLine 字段(其余配置原样不动,改前自动备份 settings.json.bak.<时间戳>)、最后自检脚本可运行。
安装后下次状态栏刷新(发一条消息或重启 Claude Code)即生效。之后改脚本本身无需重启——statusLine 每次刷新都重新执行脚本。
装到 OpenClaw 或自定义目录:
CLAUDE_HOME=~/.openclaw bash .../install.sh(前提是该 harness 支持 statusLine)。
显示含义
<模型名> <进度条> <百分比> (<已用>/<上限>),颜色三档:
- 🟢 绿
<60%充裕 🟡 黄60–85%留意,可准备 /compact 🔴 红≥85%该 /compact 了
工作原理(及三个关键坑)
脚本从 stdin 读取 Claude Code 传入的会话信息(含 transcript_path),反向扫描 transcript 取最近一次的 context 占用 = input_tokens + cache_read_input_tokens + cache_creation_input_tokens。三个容易踩的坑,脚本都处理了——这也是它存在的价值:
- 跳过全 0 的合成消息:API 出错会在 transcript 末尾留下一条
<synthetic>全 0 usage 消息。天真地取"最后一条 usage"会读到 0、误显示 0%。脚本跳过 token 全 0 的消息。 - /compact 边界(最隐蔽):
/compact会追加一条带compactMetadata的消息,其中postTokens是压缩后的真实大小——但这条消息本身没有 usage。若只取"最后一条 usage",会穿透到压缩之前的旧值,表现为"compact 完了百分比还是不降"。脚本的规则是:反向遍历时,compact 边界(postTokens)和真实 usage 谁先遇到用谁——于是 compact 后的空窗期显示postTokens(已压缩),等你发新消息后自动切换到含工具/系统提示开销的真实值。 - 1M vs 200k 上限:Opus 1M context 模型上限 100 万 token,其余 20 万。脚本按 statusLine 传入的 model id 是否含
1m自动判断;可用环境变量CLAUDE_CTX_LIMIT强制覆盖。
自定义
- 上限:
export CLAUDE_CTX_LIMIT=1000000,或改脚本context_limit()的默认值 - 颜色阈值:改脚本里
pct >= 85/pct >= 60两个数 - 进度条长度:改脚本
bar_len
文件结构
~/.claude/context-monitor/statusline-context.py # statusLine 调用的脚本
~/.claude/settings.json # 增量加了一个 statusLine 字段
~/.claude/settings.json.bak.<时间戳> # 改动前的自动备份卸载
bash ~/.claude/skills/context-monitor/scripts/uninstall.sh只移除本 skill 写入的 statusLine(会校验 command 指向本脚本才删,不误伤你别的 statusLine),并删除 ~/.claude/context-monitor/,改前同样自动备份 settings.json。
边界说明
- 仅适用于支持 statusLine 机制的 Claude Code;OpenClaw 等若无 statusLine 则不适用。
- 百分比是"逼近度"参考,不等于 autocompact 的精确触发点。它的作用是让你看见并主动决策,而不是替代 autocompact。
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

