Backfill issue -079 — The AI Toolchain

EDITIONS tail grep head diff uniq

The daily firehose — everything the toolchain shipped today, already filtered.

// HOW THIS ISSUE IS MADE

We read every release from the 174 tools on our watchlist at the source — GitHub and GitLab release notes, vendor release pages and changelogs, project blogs and feeds, vendor press releases, and the source code behind the tag. Bug-fix-only releases and non-product newsroom noise are dropped; what's left is summarized down to the new capability, how to try it, and any screenshots or videos the release itself published. Every entry links to the sources it was built from.

VIEW

ISSUE VIEW full issue

Do you prefer this view?

$ tct list   # 8 tools matched

AI & LLM Tooling

◆ AI Coding Agents

GitHub Copilot CLI

v1.0.57 NOTES

Copilot CLI v1.0.57 adds HTTP/2 opt-in, COPILOT_HOME support, showTipsOnStartup setting, and smarter MCP/LSP/plugin behaviors.

└──▷ GET THIS VERSION

$ git clone --branch v1.0.57 https://github.com/github/copilot-cli.git
# already have the repo? check out this version:
$ git checkout v1.0.57

└──▷ USE IT

Disable startup tips in a scripted or CI-adjacent interactive session by setting showTipsOnStartup in repo-level config.

json

{
  "showTipsOnStartup": false
}

›Adds COPILOT_ENABLE_HTTP2=1 environment variable to opt into HTTP/2 transport; default networking is now HTTP/1.1 for broader reliability.
›Adds COPILOT_HOME environment variable to control the server discovery registry directory.
›Adds showTipsOnStartup setting to control whether startup tips are displayed at session start.
›Adds extraKnownMarketplaces support in .github/copilot/settings.json so copilot plugin marketplace list honors repo-level marketplace registries.
›/lsp show, /lsp test, and /lsp reload now correctly discover project LSP config when the CLI is launched from a subdirectory.

+8 moreshow less

›/diff defaults to branch diff when there are no unstaged changes.
›In Azure DevOps-only repositories, the built-in GitHub MCP server now exposes the web_search tool instead of being fully disabled.
›Canvas providers can return file:// URLs in open results, enabling local file previews.
›Canceling a running shell command via Ctrl+C (on a !command or agent command, including sandboxed and background-promoted shells) now terminates the entire process tree.
›preToolUse hook errors now deny the tool call instead of silently permitting execution.
›Plugin slash commands (/plugin install, uninstall, update, marketplace add/remove/browse) now show immediate progress feedback during operations.
›Surfaces the underlying reason (e.g. GitHub API rate limit) when SDK auth-token validation fails, replacing a misleading generic error message.
›Actionable error message shown when the GitHub API rate limit is hit during copilot update.

Was this useful?

OpenAI Codex CLI

rust-v0.136.0 NOTES

Lightweight coding agent that runs in your terminal

Codex CLI v0.136.0 adds session archiving, OSC 8 clickable links, stdio app-server mode, Windows sandbox provisioning, and a standalone image generation extension.

└──▷ GET THIS VERSION

$ git clone --branch rust-v0.136.0 https://github.com/openai/codex.git
# already have the repo? check out this version:
$ git checkout rust-v0.136.0

└──▷ TRY IT

Archive a completed session from the CLI to prevent it from being resumed or forked accidentally.

$ codex archive <session-id>

Restore an archived session so it can be resumed or forked again.

$ codex unarchive <session-id>

Start the app-server in stdio mode for direct process-level integration without a network socket.

$ codex app-server --stdio

›Adds /archive slash command in the TUI and codex archive / codex unarchive CLI commands to protect sessions from accidental resume or fork until restored.
›TUI markdown now renders clickable web links via OSC 8 metadata, and cramped tables automatically reformat as readable key/value records while preserving link targets.
›Launches app-server in stdio mode with codex app-server --stdio for easier process-level integrations.
›Supports CODEX_API_KEY registration for approved OpenAI hosts in remote execution setup; remote-control websockets now use short-lived server tokens instead of ChatGPT access tokens.
›Adds alpha codex sandbox setup --elevated provisioning path for Windows admins, plus constrained requirements support for allowed Windows sandbox implementations.

+4 moreshow less

›Adds a feature-gated standalone image generation extension that runs through the native Codex image artifact completion pipeline.
›Exposes richer MCP server status information in app-server integrations.
›Adds focused diagnostics for MCP HTTP send failures.
›Adds experimental_request_user_input toggle in config.

Was this useful?

◆ Local LLM Runtimes

Jan AI Jan

v0.8.2 NOTES

Jan v0.8.2 adds AMD ROCm/HIP backend, pauseable downloads, and an automatic-update toggle

└──▷ GET THIS VERSION

$ git clone --branch v0.8.2 https://github.com/janhq/jan.git
# already have the repo? check out this version:
$ git checkout v0.8.2

›Adds HIP (AMD ROCm) backend support for llamacpp, enabling GPU-accelerated local inference on AMD hardware.
›Adds a toggle in Settings to enable or disable the automatic app update check.
›Adds copy and reveal buttons to API key input fields in the Providers settings.
›Adds pause and resume controls for model downloads from the hub panel.
›Adds a branded startup loader with phased status messages; defers MCP initialization until after first paint.

Was this useful?

llama.cpp

b9459 3 RELEASES · 2026-06-01 NOTES STABLE

Metal GLU kernels now support f16 inputs, reducing memory bandwidth on Apple Silicon.

└──▷ GET THIS VERSION

$ git clone --branch b9459 https://github.com/ggml-org/llama.cpp.git
# already have the repo? check out this version:
$ git checkout b9459

›Metal GLU kernels (geglu/swiglu) now accept f16 tensor inputs in addition to f32, loading and storing in the native tensor type to reduce memory bandwidth while keeping ALU compute in float for numerical stability.

2 more releases in this issue · 2026-06-01

b9455 NOTES STABLE

llama.cpp b9455 adds quantized KV cache support for tensor parallelism (TP) deployments.

└──▷ GET THIS VERSION

$ git clone --branch b9455 https://github.com/ggml-org/llama.cpp.git
# already have the repo? check out this version:
$ git checkout b9455

›Adds quantized KV cache support for tensor parallelism (TP) configurations, enabling reduced memory footprint in multi-GPU inference setups.

b9453 NOTES STABLE

llama.cpp b9453 adds EXAONE 4.5 model support including multimodal vision capabilities.

└──▷ GET THIS VERSION

$ git clone --branch b9453 https://github.com/ggml-org/llama.cpp.git
# already have the repo? check out this version:
$ git checkout b9453

›Adds EXAONE 4.5 model support with GQA and multimodal vision via <vision> and </vision> image boundary markers, routed through the Qwen2.5-VL-style encode path with window attention and optional mmproj input norm.
›Adds convert_hf_to_gguf support for EXAONE 4.5 mmproj export, enabling conversion of EXAONE 4.5 vision projector weights to GGUF format.

Was this useful?

◆ AI Agent Frameworks

LangChain LangGraph

1.2.3 2 RELEASES · 2026-06-01 NOTES STABLE

Build resilient agents.

LangGraph 1.2.3 adds v3 streaming support, WebSocket transport, and tool-dispatched subagent naming to RemoteGraph.

└──▷ GET THIS VERSION

$ git clone --branch 1.2.3 https://github.com/langchain-ai/langgraph.git
# already have the repo? check out this version:
$ git checkout 1.2.3

›Adds v3 streaming support to RemoteGraph, enabling the latest streaming protocol for remote graph execution.
›Wires RemoteGraph.interleave to sdk-py interleave_projections for interleaved stream output.
›Names tool-dispatched subagents via lc_agent_name for clearer agent identification in multi-agent graphs.
›Adds WebSocket stream transport to the Python SDK (sdk-py) as an alternative to SSE.
›Adds messages and tool call projections to sdk-py for structured stream consumption.

+1 moreshow less

›Adds v3 streaming primitives and SSE transport to sdk-py.

1 more release in this issue · 2026-06-01

sdk==0.4.1 NOTES STABLE

LangGraph SDK 0.4.1 adds interleave_projections stream decoder and v3 streaming support for RemoteGraph.

└──▷ GET THIS VERSION

$ git clone --branch sdk==0.4.1 https://github.com/langchain-ai/langgraph.git
# already have the repo? check out this version:
$ git checkout sdk==0.4.1

›Extracts stream decoders into a reusable module and adds interleave_projections stream decoder.
›Adds v3 streaming protocol support to RemoteGraph, enabling richer real-time output from remote graph execution.

Was this useful?

Other / Uncategorized

◆ AI OBSERVABILITY

Arize Phoenix

arize-phoenix-v16.5.0 2 RELEASES · 2026-06-01 NOTES STABLE

Phoenix 16.5.0 adds playground prompt saving, PXI agent span annotation, and prompt read/write tools for agents.

└──▷ GET THIS VERSION

$ git clone --branch arize-phoenix-v16.5.0 https://github.com/Arize-ai/phoenix.git
# already have the repo? check out this version:
$ git checkout arize-phoenix-v16.5.0

›Adds read_prompt_tools and write_prompt_tools capabilities to the PXI agent for programmatic prompt management.
›Adds annotate-spans skill to the PXI agent, enabling automated span annotation workflows.
›Adds summary argument to the PXI bash tool for UI preview of bash tool output.
›Adds save-prompt capability to the Playground UI, allowing prompts to be saved directly from the prompt editor.
›Adds rewind, fork, and copy controls for chat messages in the agent UI.

+1 moreshow less

›Seeds default sandbox configurations for local adapters.

1 more release in this issue · 2026-06-01

arize-phoenix-v16.4.0 NOTES STABLE

Phoenix v16.4.0 adds Claude Opus 4.8 support, a span annotation tool, trace debugging skill, and playground skill to the PXI agent.

└──▷ GET THIS VERSION

$ git clone --branch arize-phoenix-v16.4.0 https://github.com/Arize-ai/phoenix.git
# already have the repo? check out this version:
$ git checkout arize-phoenix-v16.4.0

›Adds Claude Opus 4.8 as a supported model in the PXI agent interface.
›Adds a span annotation tool to the PXI agents capability, enabling annotation of spans directly from the agent.
›Adds a trace debugging skill to PXI agents for investigating traces from within the agent UI.
›Adds a playground skill to PXI agents, enabling prompt playground workflows from the agent interface.
›Makes span_kind filter comparisons case-insensitive when querying spans.

+7 moreshow less

›Surfaces PXI model credential setup, allowing users to configure model credentials within the PXI interface.
›Upgrades the PXI eval harness with message history support.
›Makes PXI quick actions dynamic per page context, adapting available actions based on the current view.
›Collapses agent context pills into a deck UI in the PXI agent panel.
›Adds no-data chart overlays to charts when data is absent.
›Adds draggable and resizable floating chat panel to the PXI agents UI.
›Lifts the trace UX feature flag, making the updated trace experience generally available.

Was this useful?

Langfuse

v3.177.0 NOTES

Langfuse v3.177.0 adds a flag to disable legacy tracing FTS, an AI telemetry toggle, and claude-opus-4-8 model pricing.

└──▷ GET THIS VERSION

$ git clone --branch v3.177.0 https://github.com/langfuse/langfuse.git
# already have the repo? check out this version:
$ git checkout v3.177.0

›Adds LANGFUSE_DISABLE_LEGACY_TRACING_IO_SEARCH environment variable to allow disabling full-text search on v3 tracing tables, useful for operators tuning query performance.
›Adds a toggle in the AI features settings to control AI telemetry collection.
›Adds built-in pricing support for the claude-opus-4-8 model.

Was this useful?

◆ VECTOR DB RAG

LanceDB

python-v0.33.1-beta.0 2 RELEASES · 2026-06-01 NOTES STABLE

LanceDB python-v0.33.1-beta.0 unifies the sync create_index API with the async API and routes merge_insert through the MemWAL LSM write path.

└──▷ GET THIS VERSION

$ git clone --branch python-v0.33.1-beta.0 https://github.com/lancedb/lancedb.git
# already have the repo? check out this version:
$ git checkout python-v0.33.1-beta.0

›Unifies the sync create_index API signature to match the async create_index API, enabling consistent index-creation code across sync and async usage.
›Routes merge_insert through the MemWAL LSM write path, enabling merge-insert operations to benefit from the LSM-based write pipeline.

1 more release in this issue · 2026-06-01

v0.30.1-beta.0 NOTES STABLE

LanceDB v0.30.1-beta.0 unifies the sync create_index API with the async API and routes merge_insert through the MemWAL LSM write path.

└──▷ GET THIS VERSION

$ git clone --branch v0.30.1-beta.0 https://github.com/lancedb/lancedb.git
# already have the repo? check out this version:
$ git checkout v0.30.1-beta.0

›Unifies the synchronous create_index API signature to match the async create_index API in Python, enabling consistent usage across both execution models.
›Routes merge_insert through the MemWAL LSM write path, unlocking improved write consistency and performance for upsert workloads.

Was this useful?

// END OF TRANSMISSION

The AI Toolchain · curated daily · June 1, 2026
Releases & features from the open-source and commercial AI tools that practitioners actually run.

The AI Toolchain — issue -079, June 1, 2026