The AI Toolchain — issue -007, August 12, 2026

Claude Code is an agentic coding tool that lives in your terminal, understands your codebase, and helps you code faster by executing routine tasks, explaining complex code, and handling git workflows - all through natural language commands.

Claude Code v2.1.229 adds plugin marketplace command sources, Remote Control offline/cloud labels, VSCode session groups, and SSE keepalive for Vertex/Bedrock.

$ git clone --branch v2.1.229 https://github.com/anthropics/claude-code.git
# already have the repo? check out this version:
$ git checkout v2.1.229

$ claude remote-control --continue

• ›Adds plugin marketplace command sources: a local command (e.g. an IDE) prints the plugin directory, re-resolved each session and applied without a restart; mode: "link" uses it in place.
• ›ListAgents now marks disconnected Remote Control sessions as offline and labels cloud sessions as cloud.
• ›Adds CLAUDE_CODE_WORKFLOW_PREFIX_STAGGER_MS env var to control workflow fan-out staggering for same-prefix sibling agents; set to 0 to disable.
• ›Adds SSE keepalive pings to gateway streaming responses during long thinking pauses, preventing idle-timeout disconnects on Vertex and Bedrock upstreams.
• ›Adds server-supplied Claude Code hook support for self-hosted runner sessions, matching managed-environment behavior.

Glamourous agentic coding for all

Crush v0.89.0 adds bash syntax highlighting, ctrl+end jump-to-bottom, and session ID on exit.

$ git clone --branch v0.89.0 https://github.com/charmbracelet/crush.git
# already have the repo? check out this version:
$ git checkout v0.89.0

• ›Adds ctrl+end keybinding to jump to the bottom of the chat and enable follow mode so new messages auto-scroll.
• ›Displays a session ID splash screen on exit, so you can quickly resume the last session on next launch.
• ›Adds syntax highlighting to bash commands shown when Crush executes via the Bash tool, improving legibility.
• ›Strips redundant cd-to-project prefix from Bash tool command display to reduce visual noise.

an open source, extensible AI agent that goes beyond code suggestions - install, execute, edit, and test with any LLM

Goose v1.46.0 adds hooks with PreToolUse denial, 25+ new providers, new slash commands, and a wave of env-var controls

$ git clone --branch v1.46.0 https://github.com/aaif-goose/goose.git
# already have the repo? check out this version:
$ git checkout v1.46.0

$ export GOOSE_OAUTH_CALLBACK_PORT=9876
goose session

$ export GOOSE_MAX_TOOL_RESPONSE_SIZE=65536
goose session

• ›Adds LOCAL_WHISPER_LANGUAGE environment variable to configure the language for local Whisper-based transcription, enabling multilingual dictation.
• ›Adds GOOSE_OAUTH_CALLBACK_PORT environment variable to pin a stable OAuth redirect_uri and avoid port-collision re-auth failures.
• ›Adds GOOSE_MAX_TOOL_RESPONSE_SIZE environment variable to set a size cap on tool output passed back to the model.
• ›Adds GOOSE_FAST_MODEL environment variable consumed by ModelConfig::with_fast to designate a lightweight model for fast subtasks.
• ›Adds MAX_CODE_BLOCK_LINES environment variable to control how many lines of code are rendered in the TUI before truncation.

$ git clone --branch v1.46.0 https://github.com/block/goose.git
# already have the repo? check out this version:
$ git checkout v1.46.0

$ GOOSE_MAX_TOOL_RESPONSE_SIZE=8192 goose run --session my-audit

$ LOCAL_WHISPER_LANGUAGE=fr goose run --session francophone-session

Qwen Code desktop v0.2.1 adds an Aliyun OSS release mirror, OpenTelemetry session lifecycle alignment, Goal v3 in ACP sessions, and improved subagent and tool activity UI.

$ git clone --branch desktop-v0.2.1 https://github.com/QwenLM/qwen-code.git
# already have the repo? check out this version:
$ git checkout desktop-v0.2.1

• ›Adopts Goal v3 in ACP sessions via the CLI (feat(cli): adopt Goal v3 in ACP sessions).
• ›Adds Aliyun OSS as a release mirror for desktop downloads.
• ›Aligns session lifecycle telemetry with the OpenTelemetry standard.
• ›Improves subagent activity rows in the web-shell UI.
• ›Improves compact tool activity display in the web-shell UI.

Zed v1.15.0 adds git.diff_base for merge-base diffs, gutter.git_gutter_width, linked JSX/TSX editing, and Wayland drag-and-drop.

$ git clone --branch v1.15.0 https://github.com/zed-industries/zed.git
# already have the repo? check out this version:
$ git checkout v1.15.0

{
  "git": {
    "diff_base": "default_branch"
  }
}

{
  "gutter": {
    "git_gutter_width": 6
  }
}

[
  {
    "context": "multiple_selections",
    "bindings": {
      "ctrl-l": "editor::SelectNextMatch"
    }
  }
]

• ›Adds git.diff_base setting to choose whether editor gutters, file-status colors, and git: diff show uncommitted changes against HEAD (value "head", the default) or all current-branch changes against the merge base with the repository's default branch (value "default_branch"); also exposed as 'Diff Against Default Branch' in the editor controls menu.
• ›Adds gutter.git_gutter_width setting to configure the pixel width of Git diff indicators in the editor gutter.
• ›Adds multiple_selections key context, enabling distinct keybindings when multiple selections are active.
• ›Adds support for self-hosted Sweep Next Edit models in OpenAI-compatible edit predictions, including the sweep prompt format and infer detection for sweep-next-edit model names.
• ›Adds linked editing for custom elements in JSX and TSX files.

Framework for orchestrating role-playing, autonomous AI agents. By fostering collaborative intelligence, CrewAI empowers agents to work together seamlessly, tackling complex tasks.

CrewAI 1.15.15 adds flow outcome, duration, and human-in-the-loop signal reporting to flow telemetry.

$ git clone --branch 1.15.15 https://github.com/crewAIInc/crewAI.git
# already have the repo? check out this version:
$ git checkout 1.15.15

• ›Reports flow outcome, duration, and human-in-the-loop signals as part of flow telemetry.

PydanticAI v2.28.0 adds real-time speech-to-speech via Agent.realtime() plus a new Crusoe provider

$ git clone --branch v2.28.0 https://github.com/pydantic/pydantic-ai.git
# already have the repo? check out this version:
$ git checkout v2.28.0

• ›Adds Agent.realtime() method for real-time speech-to-speech interactions with an agent.
• ›Adds browser WebRTC and server sideband support for real-time speech-to-speech sessions.
• ›Adds Crusoe as a new LLM provider.
• ›Adds cerebras optional dependency group.

ONNX Runtime v1.29.0 adds env-var thread-pool control, NVFP4 QMoE, PagedAttention with quantized KV cache, and broad WebGPU operator expansion.

$ git clone --branch v1.29.0 https://github.com/microsoft/onnxruntime.git
# already have the repo? check out this version:
$ git checkout v1.29.0

$ export ORT_INTRA_OP_NUM_THREADS=4
export ORT_INTER_OP_NUM_THREADS=2
python my_inference_server.py

$ export ORT_DISABLE_TELEMETRY=1
python run_inference.py

• ›Adds ORT_INTRA_OP_NUM_THREADS and ORT_INTER_OP_NUM_THREADS environment variables to set default intra-op and inter-op thread-pool sizes at process startup; explicit thread settings still take precedence, and 0 preserves machine-sized defaults.
• ›Adds ORT_DISABLE_TELEMETRY=1 environment variable to disable new POSIX telemetry (now available on Linux, macOS, Android, and iOS when built with telemetry enabled) before initialization.
• ›Adds --data_shape flag to onnxruntime_perf_test for multi-shape profiling in a single run.
• ›Adds MatMulBlockQuantizedFp4Weight and MatMulBlockQuantizedFp8Weight contrib operators on the CUDA EP, with block-scaled tensor-core/GEMV decode paths, packed FP4 decode, M-tiling, and folded W8A8 activation QDQ.
• ›Adds LinearAttentionGate, GatedRMSNorm, and GatedAdd contrib operators on the CUDA EP.

llama.cpp b10369 adds pocket-tts support with 80% faster CUDA decoding and new mmproj audio config keys

$ git clone --branch b10369 https://github.com/ggml-org/llama.cpp.git
# already have the repo? check out this version:
$ git checkout b10369

• ›Adds clip.gen.audio.frames_after_eos and clip.gen.audio.pad_short_text keys to mmproj files, letting language packs tune end-of-speech padding and short-prompt padding per pack (e.g. french_24l requests 8 tail frames, english_2026-01 pads short prompts with spaces).
• ›Adds clip.gen.audio.model_variant key to mmproj files for per-pack model variant selection.
• ›Adds mtmd_gen_inp_default API function for pocket-tts input generation.
• ›Adds pocket-tts (PocketTTS) model support to the mtmd multimodal subsystem, enabling on-device TTS inference.
• ›Replaces grouped transposed convolutions in the Seanet decoder with a GEMM + col2im_1d formulation, cutting generation time per frame by 80% on CUDA and 50% on CPU.

Phoenix v20.1.0 adds JWT client assertion for OAuth2, a trace-transfer endpoint, and refined evaluation metric chart controls.

$ git clone --branch arize-phoenix-v20.1.0 https://github.com/Arize-ai/phoenix.git
# already have the repo? check out this version:
$ git checkout arize-phoenix-v20.1.0

• ›Adds POST /traces/transfer endpoint to move traces between projects or environments.
• ›Supports JWT client assertion as an authentication method for OAuth2 login flows.
• ›Refines evaluation metric chart layout and controls in the UI.

Langfuse v4.10.0 raises trace observation cap to 20k, adds full dashboard filter options, and integrates metrics API for billing.

$ git clone --branch v4.10.0 https://github.com/langfuse/langfuse.git
# already have the repo? check out this version:
$ git checkout v4.10.0

• ›Raises the observation load cap to 20,000 per trace, unlocking inspection of much larger LLM pipeline runs.
• ›Adds all filter options and value suggestions to widget and monitor forms in dashboards, enabling more precise monitoring configurations.
• ›Integrates a metrics API for billing, connecting usage measurement to billing workflows.
• ›Adds Cursor Cloud environment support for the in-app agent.
• ›Allows users to start new conversations or switch conversations in the in-app agent while another is still running.