Track API usage across Synthetic, Z.ai, and Anthropic in one place. onWatch polls every provider's quotas, stores historical data locally, and surfaces patterns, projections, and cross-provider context—so you always know where your budget stands and which provider has headroom.
$ curl -fsSL https://raw.githubusercontent.com/onllm-dev/onwatch/main/install.sh | bash
Copy
Each provider has different quotas, reset cycles, and gotchas. onWatch handles the complexity.
Provider dashboards show a snapshot. onWatch shows the story.
After a week of data, you stop guessing and start knowing when you burn quota fastest. onWatch reveals consumption rhythms that provider dashboards never show.
Extrapolates your consumption rate to the next reset boundary. Know whether you'll run out before relief arrives — and switch providers before it happens.
See every provider's headroom at a glance. When one nears its limit, you know exactly which alternative still has capacity.
Tracks every reset boundary independently — Anthropic's 5-hour and 7-day windows, Synthetic's ~5h subscription, Z.ai's daily token budget. Know when each resets and how much went unused.
Provider dashboards show a number. onWatch shows the context, history, and projections behind it.
Time-series charts with 1h, 6h, 24h, 7d, and 30d ranges for every provider. Consumption patterns emerge that a single snapshot can never reveal.
Detects when quotas reset across all providers — Anthropic's 5-hour and 7-day windows, Synthetic's subscription cycles, Z.ai's daily token budget. Logs peak usage and total delta per cycle.
Real-time countdown to each quota reset. Extrapolates your current rate to the next boundary so you know whether you'll make it — or should switch providers.
Every agent run logs peak consumption per provider. Compare sessions side-by-side, see cycle utilization trends, and get cross-provider routing recommendations.
API providers show current usage. onWatch shows everything else.
| Capability | Synthetic | Z.ai | Anthropic | onWatch |
|---|---|---|---|---|
| Current quota usage | ✓ | ✓ | ✓ | ✓ |
| Reset time visibility | ✓ | ✕ | ✓ | ✓ |
| Historical usage trends | ✕ | ✓ | ✕ | ✓ |
| Reset cycle detection & history | ✕ | ✕ | ✕ | ✓ |
| Per-cycle consumption stats | ✕ | ✕ | ✕ | ✓ |
| Usage rate & projections | ✕ | ✕ | ✕ | ✓ |
| Per-session tracking | ✕ | ✕ | ✕ | ✓ |
| Cross-provider unified view | ✕ | ✕ | ✕ | ✓ |
| Live countdown timers | ✕ | ✕ | ✕ | ✓ |
| Open source & self-hosted | ✕ | ✕ | ✕ | ✓ |
A background agent polls your API quotas, stores snapshots in SQLite, detects patterns, and serves an intelligence dashboard. That's it.
Anyone who pays for AI coding API access and wants to know where the budget goes.
Running Cline, Claude Code, or Kilo Code on a single API key? onWatch tracks your burn rate across Anthropic, Synthetic, and Z.ai so you never get throttled mid-task.
Multiple people on the same Anthropic Max plan or shared Z.ai key? A single onWatch instance gives everyone a shared dashboard with historical trends and session tracking.
Subscribed to more than one provider? The cross-provider view shows every quota side-by-side so you can route work to whichever provider has headroom.
Deploy as a ~25 MB sidecar with zero SaaS dependencies. Use the SQLite database as a Grafana data source or pipe the REST API into your existing monitoring stack.
Choose the method that works best for you. Supports macOS (ARM64, AMD64) and Linux (AMD64, ARM64).
Auto-detects your OS and architecture, downloads the binary to ~/.onwatch/, creates .env config, sets up a systemd service (Linux) or self-daemonizes (macOS), adds to PATH, and starts tracking.
Download the binary for your platform from GitHub Releases, configure your .env, and run directly.
Want to build from source? See DEVELOPMENT.md
Install with one command: curl -fsSL https://raw.githubusercontent.com/onllm-dev/onwatch/main/install.sh | bash. Add your API keys to ~/.onwatch/.env — set any combination of SYNTHETIC_API_KEY, ZAI_API_KEY, or ANTHROPIC_TOKEN. onWatch polls each configured provider every 60 seconds, stores snapshots in SQLite, and serves a dashboard at localhost:9211 with live countdowns, charts, and cross-provider views.
Yes. onWatch monitors the API provider (Synthetic, Z.ai, or Anthropic), not the coding tool. Any tool that uses these API keys—including Cline, Roo Code, Kilo Code, Claude Code, Cursor, Windsurf, and others—will have its usage tracked automatically.
Anthropic's Pro/Max plan exposes utilization percentages and reset times for five_hour and seven_day windows, plus per-model breakdowns (seven_day_sonnet, seven_day_opus). onWatch polls this data, stores historical snapshots, and adds what Anthropic doesn't show: usage trends over time, reset cycle detection, rate projections, and cross-provider context alongside Synthetic and Z.ai. Set ANTHROPIC_TOKEN in your .env or let onWatch auto-detect from Claude Code credentials.
The Both view is onWatch's cross-provider unified dashboard. It shows Synthetic, Z.ai, and Anthropic quotas side-by-side so you can compare headroom across providers at a glance. For example, if your Synthetic search quota is at 95% but Z.ai tokens are at 34%, you know to route work to Z.ai. No other tool provides this cross-provider intelligence—provider dashboards only show their own data.
No. onWatch has zero telemetry. All usage data is stored locally in a SQLite file on your machine. The only outbound network calls are to the Synthetic, Z.ai, and Anthropic quota APIs you configure. No analytics, no tracking, no cloud. The source code is fully auditable on GitHub (GPL-3.0).
onWatch idles at ~25-30 MB RAM and peaks at ~50 MB during dashboard rendering. Breakdown: Go runtime (5 MB), SQLite in-process (2 MB), HTTP server (1 MB), polling buffer (1 MB). This is lighter than a single browser tab and designed to run as a background daemon indefinitely.
Pre-built binaries are available for macOS (ARM64 and AMD64), Linux (AMD64 and ARM64), and Windows (AMD64). onWatch is written in pure Go with no CGO dependencies, so it cross-compiles cleanly to any Go-supported platform.
Every hour you run without onWatch is an hour of usage data you'll never get back. Install in under a minute, start collecting intelligence immediately. Free, open source, zero telemetry.
