The AI Toolchain — issue -031, July 19, 2026

Autonomous coding agent as an SDK, IDE extension, or CLI assistant.

Cline CLI v3.0.45 cuts install size by 55%, adds Kimi K3 model, and enriches hub status with version numbers.

$ git clone --branch cli-v3.0.45 https://github.com/cline/cline.git
# already have the repo? check out this version:
$ git checkout cli-v3.0.45

• ›Hub status output now includes version numbers.
• ›Kimi K3 is now available as a ClinePass model.
• ›Claude Code and Codex providers are now optional and loaded on demand, reducing npm i -g cline install size from ~640MB to ~285MB.
• ›Runs now retry once automatically after refreshing expired OAuth credentials.
• ›Team runs: the spawn tool is no longer exposed to teammates, and errored teammate runs now report as failed instead of completed.

$ git clone --branch sdk/sdk/v0.0.65 https://github.com/cline/cline.git
# already have the repo? check out this version:
$ git checkout sdk/sdk/v0.0.65

Qwen Code v0.20.0 adds daemon log rotation, round-labeled review findings, WebShell history replay, and the qwen3.8-max-preview model.

$ git clone --branch v0.20.0 https://github.com/QwenLM/qwen-code.git
# already have the repo? check out this version:
$ git checkout v0.20.0

• ›Adds qwen3.8-max-preview to the Token Plan model list, making it selectable for inference.
• ›Adds bounded daemon log rotation to the CLI, preventing unbounded log growth.
• ›Bakes the round label into a review finding's identity line and key, making multi-round findings uniquely addressable.
• ›Enables replay of ChatRecord history in read-only WebShell sessions.
• ›Lets users read the full plan from the exit_plan_mode confirmation dialog before accepting.

vMLX - JANGTQ Uber Compressed MLX Models - L2 Disk Cache (survives restart) + L1 Paged (super fast ttft) + Hybrid SSM Scheduler + Cont Batching + etc!

vMLX 1.6.11 adds progressive streaming across Responses, Chat Completions, and Electron, plus extended L2 cache reuse for hybrid/mixed-attention model families.

$ git clone --branch v1.6.11 https://github.com/jjang-ai/vmlx.git
# already have the repo? check out this version:
$ git checkout v1.6.11

• ›Streams post-reasoning and post-tool answers progressively across Responses, Chat Completions, and Electron UI instead of withholding output until completion.
• ›Extends architecture-typed prefix, paged, and block-disk L2 cache reuse to hybrid and mixed-attention model families.
• ›Adds media and cache routing support for Qwen, Bonsai, Gemma, Step, and MiniMax-M3 model families (MiniMax-M2.7 remains text-only).
• ›Hardens multi-turn tool continuation, terminal events, cancellation and recovery, and reasoning/content separation.
• ›Ships two separate macOS DMG builds: vMLX-1.6.11-sequoia-arm64.dmg (macosx_14_0_arm64, macOS 14.5+) and vMLX-1.6.11-tahoe-arm64.dmg (macosx_26_0_arm64).

Phoenix v19.2.0 adds Antigravity and OpenCode MCP client instructions in settings.

$ git clone --branch arize-phoenix-v19.2.0 https://github.com/Arize-ai/phoenix.git
# already have the repo? check out this version:
$ git checkout arize-phoenix-v19.2.0

• ›Adds Antigravity and OpenCode MCP client instructions to the Settings UI.
• ›Improves the server startup banner to show dev tooling info at boot.