pi-antigravity-rotator 1.3.9 → 1.4.5

package/CHANGELOG.md

@@ -2,6 +2,41 @@
 
 ## [Unreleased]
 
+## [1.4.5] - 2026-04-28
+
+### Added
+- **Extended Token Views**: Added `2h`, `4h`, `8h`, and `12h` options to the Token Usage chart. The backend now retains up to 12 hours of minute-level resolution for accurate high-fidelity zooming.
+
+### Changed
+- **Activity Heatmap Scaling**: Expanded the heatmap to cover the last 60 days. The grid is now responsive, taking up the full width of the screen without distorting cell proportions, and the Y-axis now orders hours naturally from 00 to 23.
+- **Timezone Alignment**: X-axis labels on the Token Usage chart and the Heatmap now correctly reflect the browser's local time instead of UTC.
+
+## [1.4.1] - 2026-04-28
+
+### Added
+- **Export Data**: Added `CSV` and `JSON` export buttons to the Token Usage panel on the dashboard to easily download token metrics for external reporting.
+
+## [1.4.0] - 2026-04-28
+
+### Added
+- **Savings Estimation**: Dashboard now calculates and displays estimated USD savings based on tracked token usage compared to paid API pricing.
+- **Latency Tracking**: Proxy measures Time-to-First-Byte (TTFB) and Total duration per request. Dashboard displays p50/p95 latency stats by model.
+- **Quota Forecast**: Dashboard predicts when each model's quota will run out based on real-time requests/hour burn rate.
+- **Live Dashboard (SSE)**: Dashboard now updates in real-time via Server-Sent Events, removing the need for polling or manual refreshes.
+- **Live Request Log**: Searchable, real-time mini-log in the dashboard showing the last 200 requests (Account, Model, Status, TTFB, Total duration, Tokens).
+- **Activity Heatmap**: GitHub-style contribution grid displaying request intensity per hour across the last 7 days.
+- **Enhanced Logging**: Added explicit `START` and `END` proxy logs with unique request IDs for unambiguous tracing of 503s and 429s.
+
+### Changed
+- Configurable **Antigravity Version**: Extracted hardcoded `QUOTA_USER_AGENT` into an environment variable to prevent "This version of Antigravity is no longer supported" API errors. Now uses `1.107.0` by default.
+- Per-model cooldown granularity: Accounts can now serve fallback models (e.g. Gemini) even if their primary model (e.g. Claude) is exhausted.
+- **Token Tracking Architecture**: Refactored token usage into a tiered time-series (minutes/hours/days/months) with rolling sliding windows to avoid data drops at calendar boundaries.
+
+### Fixed
+- Fixed in-flight request leak where accounts could get stuck in a "busy" state indefinitely due to deprecated `req.aborted` handlers in Node 24.
+- Fixed rate-limit logic bypassing model-specific cooldown checks after a 429 was hit.
+
+
 ## [1.3.9] - 2026-04-27
 
 ### Fixed

package/README.md

@@ -94,15 +94,14 @@ After starting the proxy, open `http://localhost:51200/dashboard` or `http://<yo
 
 The dashboard shows:
 
-- **Model Routing table** -- Which account each model (Gemini Pro, Flash, Claude) is currently routed to
-- **Account cards** sorted by total quota (highest first), flagged/disabled last:
-  - Status badge: `active`, `ready`, `cooldown`, `flagged`, `disabled`, or `error`
-  - Model badges: which models this account is currently serving
-  - Per-model quota bars with timer type (`idle`/`7d`/`5h`) and reset countdown
-  - Request counts, last used time, token status
-  - Fresh-window policy status plus a per-account override button
-  - Error messages for flagged/errored accounts
-  - Re-enable button for disabled accounts
+- **Top Status & Controls** -- Real-time routing state, uptime, requests, and PII masking toggle.
+- **Token Usage & Savings** -- Interactive chart (`1h`, `2h`, `4h`, `8h`, `12h`, `1d`, `7d`, `1m`) showing token consumption by model, with estimated USD savings and `CSV`/`JSON` export options.
+- **Activity Heatmap** -- 60-day responsive GitHub-style contribution grid showing request intensity hour by hour.
+- **Latency (p50/p95)** -- Real-time median and 95th percentile tracking for Time-to-First-Byte (TTFB) and Total Duration per model.
+- **Quota Forecast** -- Predictive modeling showing when each model's quota will run out based on the current requests/hour burn rate.
+- **Searchable Request Log** -- Live feed of the last 200 requests with exact timestamps, models, masked accounts, status codes, and latency.
+- **Account Cards** -- Sorted by total quota. Shows status (`active`, `ready`, `cooldown`, `flagged`, `disabled`), served models, quota bars with timers, and precise error messages.
+- **Operator Panels** -- "Attention Needed" summaries for quarantined accounts and a real-time event feed of rotator actions.
 
 ![Dashboard](dashboard.png)
 
@@ -219,8 +218,11 @@ The dashboard is intended to replace day-to-day `journalctl` digging for normal
 Config files (`accounts.json`, `state.json`) are stored in `~/.pi-antigravity-rotator/` by default. Override with:
 
 ```bash
-# Environment variable
+# Environment variables
 export PI_ROTATOR_DIR=/path/to/config
+export PI_ROTATOR_QUOTA_USER_AGENT="antigravity/1.107.0 darwin/arm64"
+# Optional override for Antigravity UA version used by quota fetches
+export PI_AI_ANTIGRAVITY_VERSION=1.107.0
 
 # Or CLI flag
 pi-antigravity-rotator start --config-dir /path/to/config

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "pi-antigravity-rotator",
-  "version": "1.3.9",
+  "version": "1.4.5",
   "description": "Multi-account rotation proxy for Google Antigravity with per-model routing, real-time quota tracking, and infringement detection",
   "license": "MIT",
   "type": "module",