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.
Agno v2.3.10 adds ShopifyTools for store analytics and URL Context support for Gemini streaming.
└──▷ GET THIS VERSION
$ git clone --branch v2.3.10 https://github.com/agno-agi/agno.git
# already have the repo? check out this version:$ git checkout v2.3.10
└──▷ USE IT
Analyze Shopify sales and customer data from an agent using the new ShopifyTools toolkit.
python
from agno.tools.shopify import ShopifyTools
agent = Agent(tools=[ShopifyTools()], markdown=True)
agent.print_response('What were my top-selling products last month?')
›Adds ShopifyTools toolkit to query Shopify store backends for sales analytics, customer insights, and related data.
›Adds URL Context support for Gemini streaming requests.
$ git clone --branch 0.11.0 https://github.com/browser-use/browser-use.git
# already have the repo? check out this version:$ git checkout 0.11.0
└──▷ USE IT
Run an agent with specific cloud skills to let it leverage pre-built capabilities without manual tool wiring.
python
from browser_use import Agent, ChatBrowserUse
agent = Agent(
task='Research the top 5 competitors and summarize their pricing',
skills=['skill-uuid-1', 'skill-uuid-2'],
llm=ChatBrowserUse()
)
await agent.run()
Grant the agent access to all available cloud skills when you want maximum capability without specifying individual skill UUIDs.
python
from browser_use import Agent, ChatBrowserUse
agent = Agent(
task='Book the cheapest flight from NYC to London next month',
skills=['*'],
llm=ChatBrowserUse()
)
await agent.run()
›Adds skills parameter to Agent — accepts a list of skill UUIDs (e.g. ['skill-uuid-1', 'skill-uuid-2']) or ['*'] for all skills, enabling cloud-hosted skill execution.
›Adds ChatBrowserUse LLM class, importable from browser_use, as the provider for cloud-backed agent runs.
›Adds Mistral as a supported chat provider via ChatMistral (with schema sanitization and model presets).
›Adds fallback LLM support, allowing a secondary model to be specified when the primary LLM fails.
›Adds gemini-3-pro-preview to the list of verified models.
+2 moreshow less
›Introduces ai_step to replace extract_content on agent rerun, improving mid-run content handling.
Continue 1.36.0 adds session-persistent chat history, usage metadata on assistant messages, and the Google GenAI SDK for Gemini/Vertex streaming.
└──▷ GET THIS VERSION
$ git clone --branch @continuedev/[email protected] https://github.com/continuedev/continue.git
# already have the repo? check out this version:$ git checkout @continuedev/[email protected]
›Adds session ID support to the serve command to persist chat history across runs.
›Captures and attaches usage metadata (token counts) to assistant messages.
›Switches Gemini and Vertex AI streaming responses to the google/genai SDK.
Lightweight coding agent that runs in your terminal
Codex CLI gains Skills injection, MCP in-session login, vim TUI navigation, shell snapshotting, and an experimental tui2 frontend.
└──▷ GET THIS VERSION
$ git clone --branch rust-v0.69.0 https://github.com/openai/codex.git
# already have the repo? check out this version:$ git checkout rust-v0.69.0
›Adds Skills support: explicitly selecting a skill now injects its SKILL.md content into the turn; skills load once per session and warn on unreadable files.
›Adds MCP in-session login so users can authenticate to Model Context Protocol servers without leaving the CLI session.
›Adds vim-style navigation keys (hjkl/arrow equivalents) for both option selection menus and the transcript pager.
›Adds shell snapshotting in the exec/sandbox layer for safer and more reliable command execution.
›Introduces an experimental tui2 frontend, available behind a feature flag.
+6 moreshow less
›Elevates Linux sandbox allowances to permit sendmsg/recvmsg syscalls, unblocking more networked workloads.
›Adds support for using a remote branch when performing reviews if the local branch trails, enabling review of unsynced work.
›Rate-limit warnings now display as percentage of quota remaining for clearer visibility.
›Config writes now preserve comments and key order; model field is optional to match real-world config files.
›TUI log files now strip ANSI escape codes for cleaner plaintext logs.
›Adds codex-max prompt and tools support for experimental models.
›Adds auto-retry on transient provider errors (429, 500, 502–504) using exponential backoff (2s, 4s, 8s), with TUI retry status display and Escape-to-cancel support.
›Exposes retry configuration via retry.enabled, retry.maxRetries, and retry.baseDelayMs in settings.json.
›Emits auto_retry_start and auto_retry_end events in RPC mode for programmatic retry visibility.
›HTML exports now show line number ranges (e.g., file.txt:10-20) for read tool calls when offset/limit parameters are used, with yellow highlighting.
›Adds word_diff_enabled language setting (defaults to true) to enable word-level diff highlighting inside expanded diff hunks of fewer than 5 lines.
›Adds the ability to disable direnv integration entirely via "load_direnv": "disabled" in settings.
›Adds ESLint working directory configuration via {"lsp":{"eslint":{"settings":{"workingDirectories":["./client","./server"]}}}} in settings.
›Adds File History view accessible via right-click context menu on files in the editor, project panel, or git panel, showing commit history with author, timestamp, and commit message; clicking a commit opens a file-filtered diff view.
›Adds UI for deleting Git branches.
+14 moreshow less
›Adds ctrl-w x keybinding in Vim mode to swap the active window with the next adjacent one.
›Adds a keybinding (cmd-, on macOS / ctrl-, on Windows/Linux) to open settings.json when the Settings UI is in focus.
›Adds Show Tab Bar Buttons setting to the Settings UI to control visibility of New, Split Pane, and Zoom buttons in the tab bar.
›Adds support for deleting entire agent thread history in the agent panel.
›Adds support for @mentions after typing a slash command in ACP.
›Git now checks pushRemote and pushDefault configurations before falling back to branch remote.
›Increases the Git askpass timeout from 17 to 300 seconds.
›Adds Doxygen grammar support for C/C++ files.
›Adds Tailwind suggestions and validations for the Gleam programming language.
›Adds JavaScript highlighting via YAML injections.scm for script blocks of actions/github-script.
›Improves sorting order of pyright/basedpyright code completions for Python.
›Adds file icon for the Odin programming language.
›Improves performance for large multibuffer editors, large project searches, and project diffs.
›Git blame now runs separately from the git job queue, unblocking staging, unstaging, and committing in repositories with large git histories.
SGLang Model Gateway v0.2.4 adds WASM middleware, full OpenTelemetry tracing, multimodal VL model support, and broad performance optimizations.
└──▷ GET THIS VERSION
$ git clone --branch gateway-v0.2.4 https://github.com/sgl-project/sglang.git
# already have the repo? check out this version:$ git checkout gateway-v0.2.4
›Adds WASM middleware support, enabling safe, isolated WebAssembly plugins for custom routing logic, request/response transformation, and proprietary system integration without touching core code.
›Adds OpenTelemetry distributed tracing to the HTTP router with native trace context propagation across the inference stack.
›Adds OpenTelemetry integration to the gRPC router, extending distributed tracing coverage to both HTTP and gRPC transports.
›Makes the tokenizer builder aware of environment variables such as HF_ENDPOINT for flexible model loading configuration.
›Adds a version command to the SGLang Model Gateway CLI, exposed via the Python binding.
+13 moreshow less
›Adds support for routing Vision-Language (VL) models in the gateway router.
›Adds image processors for LLaVA, Qwen2-VL, Qwen2.5-VL, Qwen3-VL, Phi3 Vision, Phi4 Vision, Llama4 Vision, and Mistral 3 multimodal models.
›Adds a workflow for routing to external model providers.
›Adds ModelCard and ProviderType support to worker metadata for richer model configuration.
›Adds audio and moderation entries to the model card.
›Adds HTTP response status code metrics to the router.
›Implements TokenGuardBody for managing and controlling token return in responses.
›Optimizes the radix tree used in cache-aware load balancing for lower routing overhead.
›Optimizes the tokenizer for dramatically reduced CPU and memory footprint.
›Supports numactl CPU and memory binding before the gateway process starts.
›Allows the token refill rate to be set to zero.
›Promotes conversation handling to first-class routing in the OpenAI router.
›Renames the Rust package and binary from sgl-router to sgl-model-gateway.
└──▷ BREAKING ON UPGRADE
!The Rust package and binary have been renamed from sgl-router to sgl-model-gateway; any scripts, container images, or CI pipelines referencing sgl-router by name will break.
llama.cpp b7348 ships a new interactive CLI with --show-timings, --image, --audio, /read, and speculative-decoding args.
└──▷ GET THIS VERSION
$ git clone --branch b7348 https://github.com/ggml-org/llama.cpp.git
# already have the repo? check out this version:$ git checkout b7348
└──▷ TRY IT
Run an interactive session with a multimodal model, passing an image and displaying per-completion timings — useful for auditing inference latency during model evaluation.
Feed a large prompt from a file inside an interactive session without pasting it manually — handy for repeatable red-team prompt injection tests.
$ /read prompt.txt
›Adds --show-timings flag to the new CLI (enabled by default) to display inference timing information after each completion.
›Adds --image and --audio flags to the new CLI for passing media inputs directly on the command line.
›Adds /read command in the interactive CLI session to read input from a file.
›Adds arguments for speculative decoding in the new CLI experience.
›Introduces a new interactive CLI experience (the old CLI is moved to llama-completion), with a loading animation, colored output (default auto), and reasoning_content support.
+1 moreshow less
›Linux releases now ship as .tar.gz archives instead of .zip.
└──▷ BREAKING ON UPGRADE
!The old llama.cpp CLI binary is replaced by the new CLI; the previous CLI is now available as llama-completion — any scripts invoking the old CLI entry point will need to be updated.
!Linux release archives are changing from .zip to .tar.gz format — deployment scripts that unzip Linux builds will break and must be updated.
llama.cpp b7342 adds tokenizer/parser support for Ministral, Mistral Large 3, and Devstral 2 models.
└──▷ GET THIS VERSION
$ git clone --branch b7342 https://github.com/ggml-org/llama.cpp.git
# already have the repo? check out this version:$ git checkout b7342
›Adds parser support for Ministral, Mistral Large 3, and Devstral 2 model families, enabling correct tokenization and prompt formatting for these models.
└──▷ BREAKING ON UPGRADE
!Linux releases will switch from .zip archives to .tar.gz archives — deployment scripts that fetch or unpack Linux builds will need to be updated.
Arize Phoenix 12.23.0 makes email optional for LDAP authentication.
└──▷ GET THIS VERSION
$ git clone --branch arize-phoenix-v12.23.0 https://github.com/Arize-ai/phoenix.git
# already have the repo? check out this version:$ git checkout arize-phoenix-v12.23.0
›Makes email optional when configuring LDAP authentication, allowing LDAP users without a mapped email attribute to authenticate successfully.
THE AI TOOLCHAIN