Heads up This site is currently under heavy development.
Subscribe Get it delivered — the daily firehose, filtered to the tools you run, plus the documentation changes vendors never announce. Compare plans →

The AI Toolchain — issue -063, June 17, 2026

THE AI TOOLCHAIN NO. -063
Tail
THE DAILY RELEASE FIREHOSE
PUBLISHED JUNE 17, 2026 · EVERY WEEKDAY
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   # 11 tools matched
AI & LLM Tooling
◆  AI Coding Agents

Anthropic Claude Code

Sources Release notes → v2.1.181 NOTES

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 v2.1.181 adds in-prompt config setting, Apple Events sandbox opt-in, and a presence-file push-notification suppressor.

└──▷ GET THIS VERSION
$ git clone --branch v2.1.181 https://github.com/anthropics/claude-code.git
# already have the repo? check out this version:
$ git checkout v2.1.181
└──▷ TRY IT
Toggle extended thinking off mid-session without leaving the prompt.
$ /config thinking=false
Suppress mobile push notifications while you're actively at your machine by pointing the env var at a lock file you create on login.
$ export CLAUDE_CLIENT_PRESENCE_FILE="$HOME/.claude_presence" && touch "$HOME/.claude_presence" && claude
  • Adds /config key=value slash command to set any configuration value directly from the prompt (e.g. /config thinking=false), works in interactive mode, -p, and Remote Control.
  • Adds sandbox.allowAppleEvents opt-in setting so sandboxed commands can send Apple Events on macOS.
  • Adds CLAUDE_CLIENT_PRESENCE_FILE environment variable: point it at a marker file to suppress mobile push notifications while at the machine.
  • Upgrades the bundled Bun runtime to v1.4.
  • Improves streaming of long paragraphs: text now appears line-by-line instead of waiting for the first line break.
+3 moreshow less
  • Improves subagent panel: idle subagents auto-hide after 30 s, list caps at 5 rows with scroll hints, and keyboard hints appear in the footer.
  • Improves MCP OAuth browser page to match Claude Code's visual style and auto-close on success.
  • Changes fullscreen mode URL opening to require Cmd+click (macOS) / Ctrl+click, matching native terminal behavior.
└──▷ BREAKING ON UPGRADE
  • !Fullscreen mode URL opening now requires Cmd+click (macOS) / Ctrl+click; plain clicks no longer open URLs.
Was this useful?

Cline

Sources Release notes → cli-v3.0.27 2 RELEASES · 2026-06-17 NOTES STABLE

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

Cline CLI v3.0.27 adds a cline skill command, a prefilled MCP install wizard, and stricter input validation.

└──▷ GET THIS VERSION
$ git clone --branch cli-v3.0.27 https://github.com/cline/cline.git
# already have the repo? check out this version:
$ git checkout cli-v3.0.27
  • New cline skill command to install and manage skills, mirroring cline plugin install and cline mcp (defaults to the Cline agent directory).
  • New prefilled MCP install wizard command for faster MCP server setup.
  • CLI now rejects unknown commands and unquoted multi-word input with a clear error instead of silently treating bad arguments as a prompt.
└──▷ BREAKING ON UPGRADE
  • !The CLI now rejects unknown commands and unquoted multi-word input with an error rather than silently treating them as a prompt — any scripts or workflows relying on that silent fallback behavior will break.
1 more release in this issue · 2026-06-17
cli-v3.0.25 NOTES STABLE

Cline v3.0.25 adds ClinePass model support, MCP server plugins with OAuth, parallel tool calls, and ranged file reads.

└──▷ GET THIS VERSION
$ git clone --branch cli-v3.0.25 https://github.com/cline/cline.git
# already have the repo? check out this version:
$ git checkout cli-v3.0.25
  • Adds ClinePass support with selectable ClinePass models in the model picker.
  • Makes model picker sections expandable for easier navigation.
  • Adds MCP server support to plugins, including OAuth authorization during plugin install.
  • Encourages parallel tool calls to speed up task execution.
  • Caps tool output for bash commands and file reads to keep large outputs within context limits.
+1 moreshow less
  • Supports ranged reads on large files to selectively pull sections into context.
Was this useful?

Block Goose

Sources Release notes → v1.38.0 NOTES

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

Goose v1.38.0 adds goose review, a hooks system, PreToolUse denial, new providers, TUI diff viewer, and major ACP session management expansions.

└──▷ GET THIS VERSION
$ git clone --branch v1.38.0 https://github.com/block/goose.git
# already have the repo? check out this version:
$ git checkout v1.38.0
└──▷ TRY IT
Run an AI-assisted code review on your local working directory before opening a PR.
$ goose review
Switch the model mid-session without restarting, e.g. to move to a faster model for a quick subtask.
$ /model
  • New goose review command for local code review from the CLI.
  • New /model slash command in CLI sessions for switching models mid-session without restarting.
  • New TUI command on goose-cli with an integrated diff viewer.
  • New hooks system enabling automation triggers around agent actions.
  • New PreToolUse denial hook to intercept and block specific tool calls before execution.
+61 moreshow less
  • New /goal command lets the agent self-evaluate completion before finishing a task.
  • New GOOSE_MAX_TOOL_RESPONSE_SIZE env var to configure tool output size limits.
  • New GOOSE_DISABLE_TOOL_CALL_SUMMARY env var to opt out of per-tool-call summaries.
  • New GOOSE_FAST_MODEL env var honored in ModelConfig::with_fast for fast-model selection.
  • New GOOSE_OAUTH_CALLBACK_PORT env var for a stable OAuth redirect URI port.
  • Configurable MAX_CODE_BLOCK_LINES via environment variable.
  • New image read tool for ingesting images during sessions.
  • Session import functionality to bring external sessions into Goose.
  • ACP session search, keyword/type filtering, and paginated session list.
  • ACP steering messages support to guide in-flight sessions.
  • ACP thinking effort config option for controlling reasoning depth.
  • ACP session system prompt setter via custom ACP method.
  • ACP secret key support at the goose serve endpoint for authenticated serving.
  • ACP methods for config extensions management.
  • ACP last message snippets available as opt-in for sessions.
  • ACP custom method to retrieve session info.
  • Slash commands (built-in, skill, recipe) exposed in ACP server.
  • Recipe discovery and execution via ACP server.
  • Scheduled recipes now support --parameters as an optional argument.
  • Quarterly scheduling option added to the recipe scheduler.
  • CLI command to list skills with their token counts.
  • Summon subagent instructions for orchestrating subagents.
  • Open-plugins generalized as skills.
  • Projects as backend sources with system prompt injection.
  • Encrypted Nostr session sharing for cross-user collaboration.
  • Unified OTLP logging schema for cross-tool detection and telemetry.
  • Canonical thinking modes standardized across providers.
  • Hugging Face OAuth support with dedicated auth tab in settings.
  • xAI SuperGrok OAuth subscription provider added.
  • Perplexity added as a declarative OpenAI-compatible provider.
  • Alibaba (Qwen via DashScope) declarative provider added.
  • Databricks AI Gateway provider added.
  • NEAR AI Cloud provider added.
  • Scaleway provider added.
  • Vercel AI Gateway provider added with opt-in leaderboard.
  • Routstr declarative provider added.
  • FuturMix provider added.
  • oMLX declarative provider added.
  • Atomic Chat added as a declarative OpenAI-compatible provider.
  • GitHub Copilot /responses API support.
  • Scholar Sidekick MCP extension added.
  • Harbor eval runner for evaluation workflows.
  • Minimal uniffi setup enabling cross-language SDK bindings.
  • Nushell terminal and completion support.
  • Linux musl CLI builds now available for static linking.
  • Windows CUDA release artifacts published.
  • Simplified Chinese (zh-CN) UI translation.
  • Worktree-aware directory switcher for multi-worktree Git workflows.
  • Chat history search in the navigation panel.
  • Group consecutive tool calls into a single summarized chain card in the UI.
  • MCP Apps inline rendering and payload hydration/replay in Goose Desktop.
  • Deep links goose://new-session and goose://resume for opening or resuming sessions.
  • Structured per-provider config block with non-destructive provider switching.
  • Anthropic stream refusals now surfaced as visible errors instead of silent failures.
  • Strip chain-of-thought markers from custom provider output.
  • Google model inventory refresh support.
  • ACP streamable HTTP spec compliance.
  • ACP session management and search in Desktop.
  • ACP images replayed on session load.
  • Login shell PATH flag for shell environment fidelity.
  • gpt-5.5 added to known models list.
Was this useful?

SST OpenCode

Sources Release notes → v1.17.8 NOTES

The open source coding agent.

OpenCode v1.17.8 adds a Home tab toggle and a faster desktop file/folder picker in the v2 layout.

└──▷ GET THIS VERSION
$ git clone --branch v1.17.8 https://github.com/sst/opencode.git
# already have the repo? check out this version:
$ git checkout v1.17.8
  • Adds a Home tab toggle in the desktop UI to quickly switch between the Home view and your last active tab.
  • New desktop file and folder picker is faster and easier to navigate in the v2 layout.
Was this useful?

Alibaba Qwen Code

Sources Release notes → v0.18.2 NOTES

Qwen Code v0.18.2 adds daemon status API, sessions list command, workflow phase-tree, and collapsible thinking blocks with timers.

└──▷ GET THIS VERSION
$ git clone --branch v0.18.2 https://github.com/QwenLM/qwen-code.git
# already have the repo? check out this version:
$ git checkout v0.18.2
└──▷ TRY IT
Check whether the Qwen Code daemon is running and inspect its reported capabilities from the CLI.
$ qwen daemon status
List recent sessions as machine-readable JSON, capped to the 10 most recent, for use in scripts or auditing active contexts.
$ qwen sessions list --json --limit 10
  • Adds daemon status API via the CLI for querying daemon health and capabilities.
  • Adds sessions list subcommand with --json and --limit flags to enumerate active sessions from the CLI.
  • Adds /workflows command with workflow metadata and phase-tree display (Workflow P4).
  • Adds imperative composer API to the web-shell for external control of text input, tag insertion, and form submission.
  • Adds transcript event exposure in the web-shell so embedders can observe turn-by-turn transcript changes.
+4 moreshow less
  • Adds support for custom footer renderer in the web-shell.
  • Adds per-turn time and token counts on the collapse seam in the web-shell, displayed below the prompt.
  • Adds collapsible thinking blocks with a duration timer in the TUI.
  • Enables delivery of web-shell mid-turn messages into the running daemon turn.
Was this useful?

Zed

Sources Release notes → v1.7.2 NOTES

Zed v1.7.2 adds agent context auto-compaction with /compact, git graph custom commands via $ZED_GIT_REF, and Add to .git/info/exclude support.

└──▷ GET THIS VERSION
$ git clone --branch v1.7.2 https://github.com/zed-industries/zed.git
# already have the repo? check out this version:
$ git checkout v1.7.2
└──▷ TRY IT
Compact the current agent context mid-thread to free up context window space without starting a new session.
$ /compact
Run a custom git command against a branch or tag directly from the git graph context menu — the clicked ref is available as $ZED_GIT_REF.
$ git checkout $ZED_GIT_REF
  • Adds /compact slash command to Zed Agent for manual context compaction, alongside automatic context compaction triggered by the agent itself.
  • Adds a git graph context menu for ref labels (branches, remote refs, tags) that runs custom git commands against the clicked ref, exposed as $ZED_GIT_REF (e.g. git checkout $ZED_GIT_REF).
  • Adds Add to .git/info/exclude option under a new Git submenu in the project panel and git panel context menus, alongside the existing Add to .gitignore action.
  • Adds default_open_behavior setting to control whether selecting a project from the recent projects menu adds it to the sidebar or opens it in a new window.
  • Adds a markdown: toggle block quote action.
+15 moreshow less
  • Adds support for surrounding selected Markdown text with ~ by typing ~.
  • Adds ability to stylize built-in Python exceptions and warnings using the type.class.builtin syntax capture.
  • Adds context window usage and cost metrics display for external agents that support it.
  • Adds the ability to delete sessions from history for external agents that support it.
  • Adds a button to the git worktree picker to remove an open worktree from the current window.
  • Detects .envrc files as Shell Script.
  • Adds an optional tree view for commit details changed-file lists.
  • Makes zed:// links clickable in the terminal.
  • Adds Chromium code review and SourceHut remote provider icons to the git panel.
  • Adds scope picker (User vs. project) to the settings window breadcrumb for sub-pages such as Skills and per-language settings.
  • Moves agent skills management into the settings UI.
  • Allows loading skills whose descriptions exceed the 1024-byte limit by showing warnings instead of failing to load them.
  • Adds OpenCode models: MiniMax M3, Qwen 3.7 Plus, DeepSeek V4 Flash, MiniMax M3 Free, and Nemotron 3 Ultra Free.
  • Improves Markdown preview styles for legibility and readability.
  • Improves outline symbol search with more consistent fuzzy matching and support for multi-term queries.
Was this useful?
◆  AI Model & Data Infrastructure

Ollama

Sources Release notes → v0.30.10 NOTES

Get up and running with Kimi-K2.6, GLM-5.2, MiniMax, DeepSeek, gpt-oss, Qwen, Gemma and other models.

Ollama v0.30.10 brings Command A and North family models to Apple Silicon via the MLX engine.

└──▷ GET THIS VERSION
$ git clone --branch v0.30.10 https://github.com/ollama/ollama.git
# already have the repo? check out this version:
$ git checkout v0.30.10
  • Supports Command A and North family models running on Apple Silicon using the MLX engine.
Was this useful?
◆  AI Agent Frameworks

Microsoft Semantic Kernel

Sources Release notes → python-1.43.1 NOTES

Semantic Kernel Python 1.43.1 adds function_choice_behavior support to Azure AI and OpenAI Assistant agents.

└──▷ GET THIS VERSION
$ git clone --branch python-1.43.1 https://github.com/microsoft/semantic-kernel.git
# already have the repo? check out this version:
$ git checkout python-1.43.1
  • Adds function_choice_behavior support to Azure AI and OpenAI Assistant agents, enabling tool-call control for agent interactions.
Was this useful?
◆  Local LLM Runtimes

llama.cpp

Sources Release notes → b9689 4 RELEASES · 2026-06-17 NOTES STABLE

Metal backend concat operator gains f16, bf16, i8, i16, and i64 tensor type support on Apple Silicon.

└──▷ GET THIS VERSION
$ git clone --branch b9689 https://github.com/ggml-org/llama.cpp.git
# already have the repo? check out this version:
$ git checkout b9689
  • Extends the Metal backend kernel_concat operator to support f16, bf16, i8, i16, and i64 tensor types in addition to the existing f32 and i32, enabling broader model architectures to run GPU-accelerated on Apple Silicon.
3 more releases in this issue · 2026-06-17
b9688 NOTES STABLE

llama.cpp b9688 adds a server-side model management API with download, delete, and SSE real-time update endpoints.

└──▷ GET THIS VERSION
$ git clone --branch b9688 https://github.com/ggml-org/llama.cpp.git
# already have the repo? check out this version:
$ git checkout b9688
  • Adds a model management API to the llama.cpp server router, including endpoints to download models, delete models, and stream real-time status updates via SSE (Server-Sent Events).
b9675 NOTES STABLE

llama.cpp b9675 extends SYCL FP16 support to SQR, SQRT, LOG, SIN, COS, and CLAMP operations.

└──▷ GET THIS VERSION
$ git clone --branch b9675 https://github.com/ggml-org/llama.cpp.git
# already have the repo? check out this version:
$ git checkout b9675
  • Enables FP16 support in SYCL backend for the SQR, SQRT, LOG, SIN, COS, and CLAMP operations, broadening half-precision GPU acceleration on Intel SYCL devices.
b9673 NOTES STABLE

llama.cpp b9673 adds GGML_SYCL_USM_SYSTEM env var to enable VRAM overcommit via USM system allocations on SYCL devices.

└──▷ GET THIS VERSION
$ git clone --branch b9673 https://github.com/ggml-org/llama.cpp.git
# already have the repo? check out this version:
$ git checkout b9673
└──▷ TRY IT
Run a model that exceeds available VRAM on a SYCL device (e.g. Intel B580) by enabling USM system allocations for automatic host/device memory migration.
$ GGML_SYCL_USM_SYSTEM=1 ./examples/sycl/test.sh -m Qwen3.5-27B-Q3_K_M.gguf -lv 4
  • Adds GGML_SYCL_USM_SYSTEM environment variable to enable optional USM system allocations for large GPU buffers (≥ 1 GB) on SYCL devices, allowing VRAM overcommit by letting the OS manage host/device memory migration; falls back to regular allocations if unsupported.
Was this useful?
Other / Uncategorized
◆  AI OBSERVABILITY

Arize Phoenix

Sources Release notes → arize-phoenix-v17.8.0 NOTES

Phoenix v17.8.0 adds an Agent GraphQL skill and theme-aware UI pre-loading.

└──▷ GET THIS VERSION
$ git clone --branch arize-phoenix-v17.8.0 https://github.com/Arize-ai/phoenix.git
# already have the repo? check out this version:
$ git checkout arize-phoenix-v17.8.0
  • Adds an Agent GraphQL skill, enabling agent-oriented queries through the GraphQL interface.
  • UI background color pre-loading now respects saved theme preferences and OS-level dark/light mode settings.
Was this useful?

Langfuse

Sources Release notes → v3.188.0 NOTES

Langfuse v3.188.0 adds rate limits to web callouts and the agent, blob storage integrations to core data S3 export, and a subquery-IN rewrite for observations.

└──▷ GET THIS VERSION
$ git clone --branch v3.188.0 https://github.com/langfuse/langfuse.git
# already have the repo? check out this version:
$ git checkout v3.188.0
  • Adds blob_storage_integrations to the core data S3 export worker, expanding what gets included in bulk exports.
  • Adds rate limits to web callouts via the feat(web-callouts) surface.
  • Adds rate limits to the agent component.
  • Introduces a subquery-IN rewrite for observations-v2 queries, available behind an environment flag.
Was this useful?
my-toolchain — 0 tools
paste an install list to detect your tools

A brew list, a Brewfile, requirements.txt, a Dockerfile — or just the product names, free-form. Nothing leaves your browser.

    browse all tools →