Track every AI subscription. Monitor every quota. Control every dollar.
Local-first. Zero daemon by default. Full provenance on every data point. MIT licensed.
Developers in 2026 spend $200-600/month across 8+ AI tools. Each has its own quota dashboard, its own billing page, its own renewal date. Quota trackers show you API usage. Finance apps show you bank transactions. Nothing shows you both.
Niyantra is a local-first dashboard that combines AI quota monitoring with subscription management, budget intelligence, and AI agent integration in a single binary. No cloud. No accounts. All data on your machine.
curl -fsSL https://raw.githubusercontent.com/bhaskarjha-com/niyantra/main/install.sh | sh# Download and extract latest release
$release = (Invoke-RestMethod "https://api.github.com/repos/bhaskarjha-com/niyantra/releases/latest").tag_name -replace '^v',''
$url = "https://github.com/bhaskarjha-com/niyantra/releases/latest/download/niyantra_${release}_windows_amd64.zip"
Invoke-WebRequest -Uri $url -OutFile "$env:TEMP\niyantra.zip"
Expand-Archive -Path "$env:TEMP\niyantra.zip" -DestinationPath "$env:LOCALAPPDATA\niyantra" -Force
# Add to PATH (run once)
$p = [Environment]::GetEnvironmentVariable("Path","User"); if ($p -notlike "*niyantra*") { [Environment]::SetEnvironmentVariable("Path","$p;$env:LOCALAPPDATA\niyantra","User") }go install github.com/bhaskarjha-com/niyantra/cmd/niyantra@latestPre-built binaries for macOS, Linux, and Windows are available on the Releases page.
git clone https://github.com/bhaskarjha-com/niyantra.git
cd niyantra
# macOS/Linux (with make)
make build
# Windows (or without make)
go build -o niyantra.exe ./cmd/niyantraDon't have Antigravity running? No problem. The demo command seeds realistic sample data so you can explore immediately:
# macOS/Linux
make demo # Seeds data + launches serve
# Windows
go run ./cmd/niyantra demo
go run ./cmd/niyantra serveWhen you're ready to use real data:
niyantra snap # Capture your Antigravity account's quota (Antigravity must be running)
niyantra serve # Dashboard shows your real data- Multi-account developers juggling work + personal AI accounts
- AI power users paying $100-500/month across multiple subscriptions
- Privacy-conscious developers who want local-only data, no cloud dashboards
- Freelancers & contractors tracking AI costs per client or project
- Anyone tired of checking 5 different dashboards to answer "am I ready to code?"
Auto-capture Antigravity per-model quotas (Claude, Gemini, GPT) with rolling 5-hour reset detection. Track your native Google AI Credits balances continuously. Monitor Codex/ChatGPT via an API proxy and track Claude Code rate limits via statusline bridge. Quick Adjust lets you fine-tune stale values with ±5%/±10% buttons right on the dashboard. See who's ready, who's exhausted, and when resets happen.
Track subscriptions across 26+ AI platforms with renewals, spending breakdowns, and CSV export. Set a monthly budget and get forecasts before you overspend. Visual renewal calendar so nothing surprises you.
Switch Advisor ranks your accounts and tells you which one to use right now. MCP Server (8 tools) lets your AI agent check quotas, analyze spending, and get routing recommendations mid-task.
SQLite database. No cloud. No telemetry. Full provenance audit trail on every snapshot -- you can always verify how and when data entered the system.
4 tabs -- Quotas, Subscriptions, Overview, Settings
| Tab | What it shows |
|---|---|
| Quotas | Provider-sectioned layout (Antigravity/Codex/Claude), per-model progress bars with reset timers, Quick Adjust (±5%/±10%), provider & status filters, split-button snap, twin-axis history chart, AI Credits tracking |
| Subscriptions | Hybrid card + provider layout with spend summary, search, 26 platform presets, CSV export |
| Overview | Monthly budget vs actual, switch advisor, provider health cards, Codex status, sessions timeline, renewal calendar |
| Settings | Auto-capture, polling interval, notifications, Claude bridge, Codex, backup/restore, command palette (Ctrl+K) |
| Command | What it does |
|---|---|
niyantra snap |
Capture current Antigravity account's quota |
niyantra status |
Show all accounts' readiness (offline) |
niyantra serve |
Launch web dashboard at localhost:9222 |
niyantra mcp |
Start MCP server (stdio) for AI agents |
niyantra demo |
Seed database with sample data |
niyantra backup |
Create timestamped database backup |
niyantra restore <file> |
Restore from backup |
niyantra version |
Print version |
Flags: --port 9222 --bind 127.0.0.1 --db ~/.niyantra/niyantra.db --auth user:pass --debug
Environment Variables: NIYANTRA_PORT NIYANTRA_BIND NIYANTRA_DB NIYANTRA_AUTH (CLI flags take precedence)
Niyantra exposes quota intelligence to AI coding agents via the Model Context Protocol.
9 tools: quota_status model_availability usage_intelligence budget_forecast best_model analyze_spending switch_recommendation codex_status quota_forecast
Add to your MCP client config (Claude Desktop, Antigravity, etc.):
{
"mcpServers": {
"niyantra": {
"command": "niyantra",
"args": ["mcp"]
}
}
}Then ask: "What's my quota?" or "Which account should I use?" or "How much am I spending?"
| Feature | Niyantra | Quota trackers (onWatch, CodexBar) | Sub trackers (Wallos) | Account managers (Antigravity-Manager) |
|---|---|---|---|---|
| AI quota monitoring | 3 providers (AG + Codex + Claude) | 9-16+ providers | — | 4+ providers |
| Multi-account per provider | ✅ 28+ (passive, safe) | ❌ Single-account | — | ✅ (active switching |
| Subscription management | 26 AI platforms, renewals, CSV | — | Generic subs | — |
| Budget forecasting | Monthly budget with projections | — | Basic budget | — |
| Switch advisor (account routing) | Multi-factor scoring engine | — | — | — |
| MCP for AI agents | 8 tools over stdio | — | — | — |
| Renewal calendar | Visual month view | — | Calendar view | — |
| Activity audit trail | Full provenance on every data point | — | — | — |
| Zero-daemon default | Manual mode, opt-in auto | Daemon by default | N/A | Daemon |
| Account switching | ❌ (by design — safety) | ❌ | — | ✅ ( |
| Docker support | Planned (Phase 14) | ✅ | ✅ | — |
| License | MIT | MIT / GPL-3 | GPL-3 | MIT |
Competitive position: Niyantra scores 24/37 features in our competitive analysis — leading all 28 tools evaluated across 8 market categories. The closest competitor (onWatch) scores 12/37. See VISION.md for full market positioning.
Niyantra detects the Antigravity language server process. Make sure:
- Antigravity IDE is running (not just installed)
- A workspace/file is open (the language server starts when you open a project)
- Try
niyantra snap --debugfor detailed detection output
Yes! Niyantra's subscription manager, budget tracking, and renewal calendar work independently. Use niyantra demo to explore, or add subscriptions manually without any Antigravity integration.
No. By default, Niyantra makes HTTP calls purely locally to 127.0.0.1 (your local Antigravity Language Server). There is no telemetry or analytics. Note: If you explicitly enable the optional Codex Capture feature, your local credentials will securely query OpenAI's authorization servers to monitor external quota balances. For detailed behavior, see SECURITY.md for the full threat model.
Re-run the install command or download the latest from Releases. Your data (~/.niyantra/niyantra.db) is preserved across updates.
| Dependency | Purpose |
|---|---|
modernc.org/sqlite |
Pure-Go SQLite -- no CGo, true single binary |
go-sdk/mcp |
Official MCP Go SDK for AI agent integration |
| Go stdlib | Everything else -- HTTP, JSON, embed, crypto |
No web frameworks. No ORMs. Chart.js bundled locally from embedded assets. Frontend: 27 TypeScript modules (strict mode) bundled by esbuild into a single IIFE.
| Document | Content |
|---|---|
| USER_GUIDE.md | Complete feature guide — start here |
| VISION.md | Product vision, market position, roadmap (Phases 1-16), competitive analysis |
| ARCHITECTURE.md | System design, data flow, security model |
| API_SPEC.md | REST API reference (31 endpoints) |
| DATA_MODEL.md | SQLite schema v11 (11 tables) |
| SECURITY.md | What data is accessed, network behavior, threat model |
| TESTING.md | Test cases |
| CONTRIBUTING.md | Development setup, code style, PR guidelines |
| CHANGELOG.md | Version history |
See CONTRIBUTING.md for development setup. Quick start:
git clone https://github.com/bhaskarjha-com/niyantra.git
cd niyantra
make build # Build binary
make test # Run tests
niyantra demo # Seed sample data
niyantra serve # Launch dashboard for developmentMIT -- (c) 2026 Bhaskar Jha