free-claude-code Releases

Added extensive features for realtime LLM services including new event handlers, metadata frames, VAD analyzers, and various service parameters

Jun 17, 2026

This release fixes critical Docker UI credential injection issues and introduces credential-aware scouting capabilities. Key highlights include fixing a 403 error on self-host, persisting telemetry for artifact metadata violations, and adding credential-typed synthesis in code-block mode.

Jun 11, 2026

Minimal release notes: "No functional changes. Added automated builds only"

Jun 11, 2026

A major refactoring focused on improving the workflow editor and tagging system, along with splitting the large Copilot tools module. Key highlights include the introduction of Phase 6.1 tagging features, the split of the 9.5k-line Copilot tools.py, and fixes for browser profile dropdown clipping.

Jun 10, 2026

This release introduces SIP REGISTER support and integrates Livekit as a new communication leg for VoiceBlender. Key highlights include the ability to call AORs via Livekit and the addition of SIP REGISTER functionality.

Jun 10, 2026

Focuses on security hardening, Copilot code-first authoring, and new frontend capabilities like speech-to-text and credential folders. Key highlights include adding speech-to-text for prompts, implementing code-first copilot policies, and fixing high-severity vulnerabilities in the langchain integration.

Jun 8, 2026

Primarily a dependency maintenance release updating the Hono web framework and Vitest test runner. Key highlights include the patch update of Hono and the major bump of Vitest in subpackages (held back in the root due to Node 18 compatibility).

Jun 8, 2026

Jun 5, 2026

Jun 4, 2026

Includes significant security patches for critical vulnerabilities and enhancements to the workflow scheduling and Copilot features. Key highlights include fixing CVE-2026-47429 and CVE-2026-44492, adding a generic remote CDP browser type, and enabling OSS workflow schedules.

Jun 4, 2026

Jun 2, 2026

A version bump introducing utility features like analytics dashboards and browser profile management improvements. Key highlights include the addition of an analytics dashboard backend, browser profile MCP auth reuse, and the replacement of a dynamic GitHub button with a static link.

May 31, 2026

The first tagged release of BarraCUDA, introducing cross-platform backends for running Triton kernels on CPU and RISC-V. Key highlights include the new CPU backend for native x86 execution, RISC-V support for QEMU, and cross-backend differential testing.

May 29, 2026

This release introduces a multi-agent framework (`pipecat.workers`) to enable peer-to-peer communication between workers. It adds a new `UIWorker` for voice agents to control web interfaces, new transport integrations like Vonage, and new LLM services including Inception. Additionally, it updates default models and improves performance for various services.

May 29, 2026

Adds a Message Queue (MoQ) endpoint to the OpenAPI specification generation for VoiceBlender. Key highlights include the addition of the MoQ endpoint to the generated API documentation.

May 26, 2026

The first official release of SenseVoice, a foundation model designed for multilingual speech understanding with ultra-fast inference. Key highlights include support for 50+ languages, speech emotion recognition, audio event detection, and a 70ms inference time for 10 seconds of audio.

May 25, 2026

A maintenance release focused on dependency version bumps for the prooflayer-scanner and scanner-lite packages, including bug fixes for the qs library and dev dependency updates for PostCSS.

May 24, 2026

May 22, 2026

A feature update for VoiceBlender introducing role-based audio routing capabilities, allowing for more granular control over audio distribution within a room.

May 21, 2026

A bug fix release for VoiceBlender ensuring that audio leg sample rates are preserved to prevent incorrect audio mixing during processing.

May 20, 2026