onWatch

Monitor quota usage across Synthetic, Z.ai, Anthropic, and GitHub Copilot from a single dashboard. onWatch is a free CLI agent that runs background polling, stores historical data in SQLite, and serves a web dashboard with consumption projections and alerts.

★★★★★

5.0(1 ratings)

• Unified multi-provider panel. onWatch aggregates quotas from Synthetic, Z.ai, Anthropic, and GitHub Copilot in a single view. It displays side-by-side utilization, time until reset, and projections for each provider, allowing quick comparison of which service has available capacity.
• Automatic reset cycle detection. The program autonomously identifies the different periods for each API: Anthropic's 5-hour and 7-day windows, Synthetic subscription cycles, Z.ai daily token limits, and GitHub Copilot's monthly cycle. It records peak usage and the total delta for each cycle.
• Consumption projections and countdown. Extrapolates the current consumption rate until the next limit reset. Shows in real-time if quota will be exhausted before renewal, facilitating the decision to switch providers before being rate-limited.
• Historical usage trends. Stores quota snapshots every 60 seconds and generates time-series charts with ranges of 1 hour, 6 hours, 24 hours, 7 days, and 30 days. Reveals consumption patterns that native provider dashboards do not show.
• Per-session tracking. Each execution of agents like Cline, Roo Code, Claude Code, or Cursor logs the peak consumption per provider. Enables comparing sessions and viewing per-cycle usage trends to optimize routing between APIs.
• Email alerts and push notifications. Configure warning or critical thresholds per quota. Receive alerts via SMTP email (with AES-256 encrypted passwords) or push notifications via PWA using the Web Push protocol (VAPID) without external services.
• Self-updating binary. onWatch can update itself from the CLI or the dashboard with a single click. On Linux, it automatically restarts the systemd service; on macOS, it replaces the binary in a controlled manner without service interruption.
• Installable as PWA. The dashboard can be installed from the browser as a native app, functioning on desktop and mobile. Push notifications require HTTPS, managed locally via auto-generated VAPID keys.
• Auto-refresh of Anthropic OAuth tokens. Automatically renews Anthropic tokens before expiration. Detects credential changes in Claude Code and resumes polling when new tokens appear, without manual intervention.
• REST API and Docker support. Exposes over 15 JSON endpoints for integration with monitoring systems like Grafana. Provides a distroless Docker image (~12 MB) with a non-root user and a ready-to-use docker-compose for DevOps environments.
• GitHub Copilot quota detection. Monitors the monthly limit for premium_interactions (300-1500), as well as unlimited chat and completions. Uses GitHub PAT with the copilot scope and automatically detects the monthly reset cycle.
• Z.ai data normalization. Interprets and unifies Z.ai's confusing field names (usage as limit, currentValue as actual usage) to clearly present the daily limit for tokens, time, and tool calls.
• Efficient Synthetic polling. Queries the /v2/quotas endpoint every 60 seconds without consuming quota. Breaks down the subscription (~5h), search.hourly (250 req/h), and toolCallDiscounts (~24h).
• Enterprise-level security. Implements AES-256-GCM encryption for SMTP passwords, constant-time authentication, parameterized SQLite queries, and zero telemetry. All data remains on the local machine.
• No-waste mode near reset. Shows how much quota went unused in each cycle and the exact time until the next reset, preventing capacity waste when a limit is about to renew.

Development of onWatch began in 2024 by the onLLM.dev team to solve the lack of historical and comparative visibility among major AI API providers. The tool is written entirely in Go 1.25+ with no CGO dependencies, enabling clean cross-compilation for macOS (ARM64/AMD64), Linux (ARM64/AMD64), and Windows (AMD64). The core of its data intelligence resides in a local SQLite database that stores consumption history, allowing pattern analysis without relying on cloud services.

Alternatives to onWatch:

Scanframe

Scanframe is a desktop application for extracting text from video files using OCR technology.

Price: Free   Size: 407 MB   Version: 1.1.1   OS: Windows

MarkdownShellExtensions

MarkdownShellExtensions is an application that integrates editing and preview functions for Markdown files into Windows Explorer.

Price: Free   Size: 35 MB   Version: 2.5.0   OS: Windows

Markdown Help Viewer

Integrated help system based on Markdown and HTML format files for Delphi and Windows applications.

Price: Free   Size: 24.4 MB   Version: 2.3.6   OS: Windows