@anonympins/fingerprint 0.6.4 → 0.6.6

package/CHANGELOG.md

@@ -1,3 +1,32 @@
+## Version 0.6.6
+
+- 🛡️ **Medium Threshold, Subnet Score & Decay Persistence Fixes**:
+  - Resolved an issue where restrictive actions were not properly applied when crossing the medium threshold (`medium_threshold`).
+  - Improved and corrected the subnet reputation calculation (`subnetScore`): anomaly metrics are now anchored to a stable, unique hardware identifier (`stable_fp_id`). This neutralizes bypass tactics like cookie dropping without penalizing legitimate users sharing the same network block. (Contributed by @anonympins)
+  - **Fixed Sliding Temporal Decay Persistence**: Corrected the subnet metrics decay calculation across PHP, Java, and Python versions, ensuring that decayed metrics are immediately persisted in the active data store (`IStore`) instead of being lost on subsequent evaluations, preventing old activity spikes from overwriting calm history. (Contributed by @anonympins)
+
+- 🌐 **Protocol Anomaly Scoring**:
+  - Integrated a comprehensive protocol anomaly evaluation module (`protocolAnomalyScore`) that jointly analyzes low-level signatures (TLS handshakes, HTTP/2 stream settings, and QUIC/HTTP/3 behaviors) to detect transport mismatches common in automated browsers and scraping tools. (Contributed by @anonympins)
+
+  - 🧠 **Machine Learning uPoW Tasks (JS/Node.js Only)**:
+    - Integrated distributed model training and evaluation tasks (`tfjs_learning` via TensorFlow.js) into the Useful Proof-of-Work (uPoW) pipeline. This allows utilizing client GPU/CPU resources for federated learning optimization, while protecting against data poisoning using gradient clipping verification. (Contributed by @anonympins)
+
+  - 🐍 **Python Engine Parity & Useful Work (uPoW) Hardening**:
+    - Integrated a robust, fully asynchronous `ProblemManager` matching the JavaScript implementation, featuring dynamic payload initializers (`generate:randomPoints`, `generate:randomAssets`) and a modular `FunctionRegistry`.
+    - Implemented strict anti-poisoning, size constraints validation, and sampling-based verification (with dynamic cheat detection) inside Python's solution integration pipeline.
+    - Hardened the `portfolio_calculate_metrics` logic to safely handle optional volatility thresholds without triggering false-positive cheating flags when `maxVolatility` is omitted.
+
+## Version 0.6.5
+
+- 🧬 **Auto-Tuner Upgrades & Threat Profiles**:
+  - **Threat-Profile Aware Tuning**: Integrated predefined security and threat profiles (e.g., `balanced`, `strict`, `api`) directly into the Auto-Tuner loop, ensuring optimized parameters dynamically align with the active defense posture.
+  - **Threshold Auto-Tuning Stabilization**: Disabled raw suspicion score threshold autotuning to prevent boundary drift, keeping decision limits highly predictable while focusing optimizer learning cycles on feature weights.
+- 🌐 **Federated Consensus & Proof-of-Work Verification**:
+  - **Federated Consensus Protocols (`federationConsensus`)**: Hardened decentralized consensus validation across federated peer nodes, improving consensus routing speed and securing verification against distributed proxy and replay tactics.
+  - **Cryptographic Sample Derivation (`deriveSampleIndices`)**: Reinforced the verification pipeline by binding sample coordinate indexing inside cryptographic challenges to a secure, time-windowed HMAC of the client IP and secret key, completely neutralizing lookup bypasses.
+- 🛡️ **Library Security Hardening**:
+  - Resolved critical security alerts and code-quality edge cases inside the main runtime library (`src/js/library.js`).
+
 ## Version 0.6.4
 
 - 🔑 **IP and TLS Session Binding in CPU Challenge**: Reinforced CPU challenge security by cryptographically binding the client's IP and `tlsSessionId` to the challenge's binary block (`baseBlock`). This effectively prevents replay attacks and the sharing of calculation solutions between distinct sessions or IP addresses.

package/README.md

@@ -77,12 +77,12 @@ npm install @anonympins/fingerprint
 ### PHP
 
 ```bash
-composer require anonympins/fingerprint
+composer install
 ```
 
 ### Python
 ```bash
-pip install fingerprint-engine
+pip install -e .
 ```
 
 ## Documentation

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@anonympins/fingerprint",
-  "version": "0.6.4",
+  "version": "0.6.6",
   "description": "Advanced anti-bot library for Node.js using multi-layer fingerprinting (JA3, client-side, headers), behavioral analysis, and adaptive Proof-of-Work (PoW) challenges to mitigate scraping, scalping, and automated threats.",
   "main": "index.js",
   "type": "module",
@@ -100,5 +100,8 @@
     "sqlite3": {
       "optional": true
     }
+  },
+  "dependencies": {
+    "@tensorflow/tfjs": "^4.22.0"
   }
 }