Track AI API quotas across Synthetic, Z.ai, Anthropic (Claude Code), Codex, GitHub Copilot & Antigravity in real time. Lightweight background daemon (<50MB RAM), SQLite storage, Material Design 3 dashboard. Zero telemetry.
API Integrationai-api-monitoringantigravityapi-monitoringclaude-codecodex
Installation
# Add to your Claude Code skills
git clone https://github.com/onllm-dev/onWatch
README.md
onWatch
Free, open-source AI API quota monitoring for developers.
Track usage across Synthetic, Z.ai, Anthropic, Codex, GitHub Copilot, MiniMax, Gemini CLI, Cursor, and Antigravity in one place.
See history, get alerts, and open a local web dashboard before you hit throttling or run over budget. Additionally, you can ingest local telemetry from your own API-driven workflows with API Integrations, keeping track of token use and spending across multiple providers.
The agent harness performance optimization system. Skills, instincts, memory, security, and research-first development for Claude Code, Codex, Opencode, Cursor and beyond.
onWatch fills the gap between "current usage snapshot" and the historical, per-cycle, cross-session view that developers actually need. It runs as a lightweight background agent (<50 MB RAM with all nine providers polling in parallel), stores historical data in SQLite, and serves a Material Design 3 web dashboard with dark/light mode.
It works with any tool that uses Synthetic, Z.ai, Anthropic, Codex, GitHub Copilot, MiniMax, Gemini CLI, Cursor, or Antigravity API keys, including Cline, Roo Code, Kilo Code, Claude Code, Codex CLI, Cursor, GitHub Copilot, MiniMax Coding Plan, Antigravity, and others.
Zero telemetry. Single binary. All data stays on your machine.
Beta: onWatch is currently in active development. Features and APIs may change as we refine the product.
If onWatch helps you track your AI spending, consider giving it a star. It helps others discover the project.
This downloads the binary to ~/.onwatch/, creates a .env config, sets up a systemd service (Linux) or self-daemonizes (macOS), and adds onwatch to your PATH.
On macOS, the installer downloads the standard binary with menubar support.
Homebrew (macOS & Linux)
brew install onllm-dev/tap/onwatch
onwatch setup # Interactive setup wizard for API keys and config
Download binaries from the Releases page. Binaries are available for macOS (ARM64, AMD64), Linux (AMD64, ARM64), and Windows (AMD64).
Or build from source (requires Go 1.25+):
git clone https://github.com/onllm-dev/onwatch.git && cd onwatch
cp .env.example .env # then add your API keys
./app.sh --build && ./onwatch --debug # or: make build && ./onwatch --debug
Or use Docker (requires Docker or Docker Compose):
cp .env.docker.example .env # add your API keys
docker-compose up -d
Or via app.sh:
./app.sh --docker --run
The Docker image uses a distroless base (~10-12 MB) and runs as non-root. An Alpine variant with shell access is also available (ghcr.io/onllm-dev/onwatch:alpine). Data persists via volume mount at /data. Logs go to stdout (docker logs -f onwatch). See Docker Deployment for details.
Configure
Edit ~/.onwatch/.env (or .env in the project directory if built from source):
SYNTHETIC_API_KEY=syn_your_key_here # https://synthetic.new/settings/api
ZAI_API_KEY=your_zai_key_here # https://www.z.ai/api-keys
ANTHROPIC_TOKEN=your_token_here # Auto-detected from Claude Code credentials
CODEX_TOKEN=your_token_here # Recommended for Codex-only setups
COPILOT_TOKEN=ghp_your_token_here # GitHub PAT with copilot scope (Beta)
ONWATCH_ADMIN_USER=admin
ONWATCH_ADMIN_PASS=changeme
At least one provider key is required. Configure any combination to track them in parallel. Anthropic tokens are auto-detected from Claude Code credentials (macOS Keychain, Linux keyring, or ~/.claude/.credentials.json). For Codex-only setups, set CODEX_TOKEN in .env; during runtime onWatch re-reads Codex auth state from ~/.codex/auth.json (or CODEX_HOME/auth.json) and picks up token changes. Copilot tokens require a GitHub Personal Access Token (classic) with the copilot scope.
onwatch # start in background (daemonizes, logs to ~/.onwatch/data/.onwatch.log)
onwatch --debug # foreground mode, logs to stdout
onwatch stop # stop the running instance
onwatch status # check if running
Open http://localhost:9211 and log in with your .env credentials.
Dashboard -- Material Design 3 with dark/light mode (auto-detects system preference). Provider tabs appear for each configured provider:
Synthetic -- Subscription, Search, and Tool Call quota cards
Z.ai -- Tokens, Time, and Tool Call quota cards
Anthropic -- Dynamic quota cards (5-Hour, 7-Day, 7-Day Sonnet, Monthly, etc.) with utilization percentages, OAuth token auto-refresh, and automatic rate limit bypass via token rotation
Codex -- Dynamic quota cards (LLMs, Review Requests) with OAuth auth-state refresh, historical cycle analytics, and multi-account support (Beta) for tracking multiple ChatGPT accounts
GitHub Copilot (Beta) -- Premium Interactions, Chat, and Completions quota cards with monthly reset tracking
MiniMax Coding Plan -- Shared quota pool tracking for M2, M2.1, and M2.5 models with 5-hour rolling window reset cycles and multi-account support for tracking multiple MiniMax subscriptions via the dashboard UI
Gemini CLI (Beta) -- Per-model quota tracking for Gemini 2.5/3.x Pro, Flash, and Flash Lite models with 24-hour reset cycles
Antigravity -- Multi-model quota cards (Claude, Gemini, GPT) with grouped quota pools, logging history, and cycle overview
Cursor -- Individual, Team, and Enterprise account tracking with auto-detected credentials from Cursor Desktop SQLite or macOS Keychain/Linux keyring, OAuth token auto-refresh, burn rate forecasts, and on-demand spend tracking
API Integrations -- Local JSONL ingestion for custom API-driven workflows and automations. Track per-integration token volume, request counts, recent activ
A Claude Code plugin that automatically captures everything Claude does during your coding sessions, compresses it with AI (using Claude's agent-sdk), and injects relevant context back into future sessions.
