CCLimitPing

README

CCLimitPing (limitping)

English | 中文

Keep your Claude Code, Codex, and GLM (Zhipu / Z.ai Coding Plan) rate-limit windows back-to-back.

These providers bill on a 5-hour rolling window (plus a weekly cap), and the 5h window starts on your first message. If you don't send anything right when a window resets, that gap is wasted — the next window only starts whenever you happen to use the tool again, drifting out of sync with your day.

limitping watches each provider and, the moment a 5h window resets, sends one minimal message to start the next window immediately — so your windows stay continuous and predictable.

claude  ✓ pinged (6.6s, 33,129 tok (in 32,956 / out 173), $0.0042)
codex   ✓ pinged (13.6s, 16,862 tok (in 16,814 / out 48), $0.0098)

Supported providers

Provider	Read usage (zero-quota)	Trigger	Auth
Claude Code	…/api/oauth/usage	claude -p	OAuth (Keychain / ~/.claude)
Codex	…/backend-api/wham/usage	codex exec	OAuth (~/.codex/auth.json)
GLM (Zhipu / Z.ai)	…/api/monitor/usage/quota/limit	minimal chat completion	API key (config / env)

[!NOTE] GLM is off by default and not yet verified on a live plan — see GLM before enabling it.

How it works

Two cleanly separated jobs:

Job	Mechanism	Cost
Trigger a new window	the official CLI (claude -p / codex exec), or a minimal API call (GLM)	a tiny slice of quota (this is the point)
Read usage & reset times	zero-quota usage endpoints (the same ones CodexBar / community plugins use)	none — never starts a window

• Claude: reads GET https://api.anthropic.com/api/oauth/usage using the OAuth token from the macOS Keychain (Claude Code-credentials) or ~/.claude/.credentials.json.
• Codex: reads GET https://chatgpt.com/backend-api/wham/usage using the OAuth token from ~/.codex/auth.json.
• GLM: reads GET …/api/monitor/usage/quota/limit (on api.z.ai or open.bigmodel.cn) using your Coding Plan API key. GLM has no standalone CLI, so the trigger is a direct minimal chat completion to …/api/coding/paas/v4/chat/completions rather than a shell-out.

Claude/Codex tokens are reused from the official tools (no separate login) and refreshed on 401. GLM uses a static API key (from config or env) — see below.

Install

limitping ships as a single self-contained binary — no Go required.

One-line script (macOS / Linux):

curl -fsSL https://raw.githubusercontent.com/wavever/CCLimitPing/main/install.sh | sh

Downloads the right prebuilt binary from the latest release into /usr/local/bin (or ~/.local/bin). Override with LIMITPING_INSTALL_DIR.

Manual download — grab the archive for your platform from the Releases page (.tar.gz for macOS/Linux, .zip for Windows):

tar -xzf limitping_darwin_arm64.tar.gz
sudo mv limitping /usr/local/bin/

Homebrew (macOS / Linux) — brew install wavever/tap/limitping (works once the Homebrew tap is set up — see .goreleaser.yaml).

From source (developers, needs Go 1.25+):

go install github.com/wavever/CCLimitPing/cmd/limitping@latest
# or, from a clone:
go build -o bin/limitping ./cmd/limitping

Each provider you enable needs its own credentials: the claude / codex CLIs logged in (Claude / Codex), or a Coding Plan API key (GLM).

Usage

limitping config init          # write ~/.config/limitping/config.toml
limitping status               # show 5h/weekly % + reset countdowns (no quota)
limitping status -v            # also print raw JSON
limitping ping                 # trigger all enabled providers now
limitping ping claude          # Claude only
limitping ping codex           # Codex only
limitping ping glm             # GLM only
limitping ping --dry-run       # show the commands without sending
limitping watch                # foreground daemon: ping each window at reset
limitping watch --dry-run      # log when pings would fire, without sending

ping shows the exact command, a live timer (a spinner on a terminal), the token usage the ping consumed (parsed from claude --output-format json / codex --json / the GLM API response), and a USD cost where available:

claude  → claude -p . --model haiku --output-format json --max-turns 1
claude  ✓ pinged (6.6s, 33,129 tok (in 32,956 / out 173), $0.0042)
codex   → codex exec --skip-git-repo-check --json -c model_reasoning_effort=low -m gpt-5.4-mini ok
codex   ✓ pinged (13.6s, 16,862 tok (in 16,814 / out 48), $0.0098)

Cost sources:

• Claude reports total_cost_usd directly.
• Codex (subscription) doesn't return a USD cost, so — like CodexBar/ccusage — we derive the equivalent API-rate cost from the LiteLLM pricing dataset (cost = non-cached-input × input + cached-input × cache-read + output × output). The dataset is cached at ~/.config/limitping/litellm_prices.json (24h TTL), with model-alias/date-suffix fallbacks. Requires [codex].model to be set so the rate can be looked up.
• GLM is a per-prompt subscription, so no per-call USD cost is shown — only the token count.

Most of Claude's input tokens are the cached Claude Code system prompt (cache_read), which is the unavoidable floor cost of claude -p.

Example status:

claude
  5h     [█████░░░░░]  51.0%  resets in 3h14m    (Sun 00:10)
  weekly [█████░░░░░]  54.0%  resets in 7h04m    (Sun 04:00)

codex (plus)
  5h     [██░░░░░░░░]  24.0%  resets in 3h15m    (Sun 00:11)
  weekly [████░░░░░░]  37.0%  resets in 111h57m  (Thu 12:53)

Configuration

~/.config/limitping/config.toml (honors $XDG_CONFIG_HOME):

weekly_threshold = 0.99   # skip pinging when weekly usage >= this (0..1), until weekly reset
reset_buffer     = "10s"  # wait this long after a reset before pinging (ensures rollover)
notify           = true   # macOS notifications on ping/skip/failure

[claude]
enabled    = true
prompt     = "."
model      = "haiku"      # cheapest tier; triggering doesn't need a SOTA model
extra_args = ["--max-turns", "1"]
align_start = ""          # optional RFC3339 anchor for the first window; empty = start ASAP

[codex]
enabled          = true
prompt           = "ok"
model            = "gpt-5.4-mini"  # cheapest Codex model for triggering
reasoning_effort = "low"  # "minimal" is rejected when web_search/image_gen tools are enabled
extra_args       = []
align_start      = ""

[glm]
enabled  = false          # opt-in: enable once you have a plan + API key
prompt   = "ok"
model    = "glm-4.6"      # cheapest standard model; flagship GLM-5/5.1 cost a multiplier
platform = "global"       # "global" = api.z.ai, "cn" = open.bigmodel.cn (Zhipu)
api_key  = ""             # empty = read from $ZAI_API_KEY (global) / $ZHIPU_API_KEY (cn)
align_start = ""

Top-level keys:

• weekly_threshold — when the weekly window is at/above this, watch stops pinging and waits for the weekly reset (unless usable credits exist).
• reset_buffer — how long to wait after a window's reset time before pinging, so the window has definitely rolled over.
• align_start (per provider) — pin the phase of your windows: set to a future RFC3339 time to delay the very first ping until then; afterwards windows chain automatically every ~5h.

Why a cheap model

Triggering a window doesn't depend on the model — any billable request starts the 5h clock — so the ping uses each provider's cheapest model to eat the least of your budget:

• Claude → haiku: also avoids the separate weekly Opus bucket.
• Codex → gpt-5.4-mini: the mini variant (see ~/.codex/models_cache.json for what your plan offers).
• GLM → glm-4.6: a standard model; the flagship GLM-5/5.1 deduct quota at a 2–3× multiplier, so avoid them for a mere trigger.

Claude/Codex don't expose per-model prices at runtime (Anthropic's local cost cache is empty; Codex's model cache has no price field), so the cheapest model is a sensible default rather than a live price lookup. Override model per provider if you prefer.

GLM (Zhipu / Z.ai Coding Plan)

GLM uses the same 5h + weekly structure as Claude/Codex, but two things differ:

• Auth is a static API key, not OAuth. Put it in [glm].api_key, or leave it empty and export ZAI_API_KEY (global) / ZHIPU_API_KEY (CN). Usage reads hit …/api/monitor/usage/quota/limit; the key goes in the Authorization header without a Bearer prefix (that's how the endpoint expects it).
• The trigger is a direct API call, because GLM has no standalone CLI. It sends a one-token chat completion to …/api/coding/paas/v4/chat/completions.

[!WARNING] Unverified on a live plan. GLM is off by default. The endpoint shapes come from community plugins; confirm on your own plan that (a) the monitor endpoint returns your real 5h/weekly windows and (b) the 5h window is anchored to your first message (so pinging at reset actually fills a gap). If GLM's window is a fixed clock or per-request sliding window, the ping won't help.

Run watch in the background (macOS, optional)

watch runs in the foreground. To keep it running via launchd, create ~/Library/LaunchAgents/com.limitping.watch.plist:

<?xml version="1.0" encoding="UTF-8"?>
<!DOCTYPE plist PUBLIC "-//Apple//DTD PLIST 1.0//EN" "http://www.apple.com/DTDs/PropertyList-1.0.dtd">
<plist version="1.0">
<dict>
  <key>Label</key><string>com.limitping.watch</string>
  <key>ProgramArguments</key>
  <array>
    <string>/ABSOLUTE/PATH/TO/limitping</string>
    <string>watch</string>
  </array>
  <key>RunAtLoad</key><true/>
  <key>KeepAlive</key><true/>
  <key>StandardOutPath</key><string>/tmp/limitping.log</string>
  <key>StandardErrorPath</key><string>/tmp/limitping.err</string>
</dict>
</plist>

launchctl load ~/Library/LaunchAgents/com.limitping.watch.plist

Cost & caveats

• Triggering consumes a little quota (~one ping per 5h ≈ 33/week). The ping uses a minimal prompt and low reasoning, so the cost is tiny but non-zero.
• The usage endpoints are unofficial and could change; they're read-only and isolated per provider for easy patching.
• macOS-first: Keychain reads and notifications are macOS-only. Codex auth.json is cross-platform; Claude on Linux uses ~/.claude/.credentials.json; notifications are a no-op off macOS.

Layout

cmd/limitping            CLI entry
internal/config          TOML config
internal/usage           normalized usage model
internal/auth            Claude (Keychain) + Codex (auth.json) tokens; GLM API key
internal/provider        per-provider ReadUsage (endpoint) + Trigger (CLI / API)
internal/pricing         LiteLLM-based USD cost lookup (Codex)
internal/scheduler       the watch engine (sleep-until-reset, weekly-respect, backoff)
internal/notify          macOS osascript notifications
internal/cli             cobra commands: status, ping, watch, config, version

Contributing

Issues and PRs are welcome. Before submitting:

gofmt -l .        # should print nothing
go build ./...
go vet ./...

Providers are isolated in internal/provider (one file each) with a small Provider interface (ReadUsage + Trigger), so adding a new provider is mostly a self-contained file plus wiring in internal/cli and internal/config.

Releasing is automated: push a tag and GitHub Actions runs GoReleaser to build the cross-platform binaries and publish a Release.

git tag v0.1.0 && git push origin v0.1.0

License

MIT © wavever