VulnClaw Releases

Replaced the HuggingFace tokenizer with a faster, lightweight alternative to improve initialization speed.

Mar 23, 2026

Focused on advanced NLP capabilities, international recognition expansion, and security fixes, including the addition of GPU support and new medical/entity recognizers.

Mar 18, 2026

This release expands API support for the default conversation mode and updates memory compilation behavior to ensure system messages are persisted immediately upon creation.

Mar 4, 2026

Added robust proxy features including mTLS support, subrequest piping, and service dependencies.

Mar 2, 2026

Minimal release notes: "## What's Changed * chore: bump version 0.16.5 by @carenthomas in https://github.com/letta-ai/letta/pull/3202 **Fu"

Feb 24, 2026

Significant expansion of PII coverage (Korean, Thai, Indian) and infrastructure improvements including full Python 3.13 support and Docker security hardening.

Feb 12, 2026

Introduces a virtual L4 stream abstraction and TLS context extensibility, alongside new filtering and proxy customization capabilities.

Jan 30, 2026

A minor maintenance release involving template updates and version bumping to prepare for subsequent feature releases.

Jan 29, 2026

Introduced enterprise-grade integrations (AHDS, Azure OpenAI) and advanced PII recognizers (Korean RRN) with improved validation and GPU support.

Sep 9, 2025

This release focuses on security hardening against HTTP/2 vulnerabilities and introduces advanced configuration options for HTTP servers and proxies, including configurable H2 options and improved lifecycle tracking. It also enhances caching mechanisms with atomic disk saving and multipart range request support while fixing numerous connection and timeout handling bugs.

Aug 15, 2025

This release introduces new hooks for upstream TCP connections, including retry logic and user timeout configuration, along with improvements to HTTP/2 health checks and error message customization. It also fixes critical deadlock issues, compilation errors on Windows, and refines HTTP/1 and HTTP/2 stream handling.

May 9, 2025

This release introduces significant additions to the framework's capabilities, including Convolutional layers (Conv1D, Conv2d) and a Simple Pooling implementation. It also adds support for the Mistral model, improves performance via f16 matmul optimizations, and implements Q8 Weight Quantization.

Mar 1, 2024

This is the initial release of Luminal, a deep learning framework focused on static computation and aggressive optimization. It serves as the first functional version, allowing the execution of the Llama 1 7B model to generate valid text.

Aug 11, 2023