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 v2.0.9 adds MCP Toolbox for Databases, Ollama Cloud support, bulk DB writes, and session_state in AgentOS run endpoints.
└──▷ GET THIS VERSION
$ git clone --branch v2.0.9 https://github.com/agno-agi/agno.git
# already have the repo? check out this version:$ git checkout v2.0.9
›Adds session_state and dependencies parameters to AgentOS run endpoints, enabling callers to pass session context directly into agent runs.
›New MCP Toolbox toolkit integrates Google's MCP Toolbox for Databases, giving agents structured access to database tools via the MCP protocol.
›Ollama Model class now supports Ollama Cloud via an API key, enabling cloud-hosted Ollama model inference alongside local deployments.
›All database implementations now support bulk writes, allowing multiple Sessions and Memories to be persisted in a single DB call.
›Workflows can now be used in AgentOS together with the Slack interface.
+2 moreshow less
›New methods added to the Scrape Graph Toolkit.
›Storage layer now raises critical errors on read/write failures rather than silently logging them, surfacing DB issues to callers.
└──▷ BREAKING ON UPGRADE
!Storage errors that were previously logged silently are now raised as exceptions; code that relied on silent failure on DB read/write errors will now encounter raised exceptions.
$ git clone --branch langchain==1.0.0a9 https://github.com/langchain-ai/langchain.git
# already have the repo? check out this version:$ git checkout langchain==1.0.0a9
›Adds create_agent function (renamed from create_react_agent) as the revamped entry point for building agents in langchain v1.
›Adds middleware support in create_agent, including a new decorator pattern for dynamically generated middleware.
›Adds dynamic system prompt middleware for runtime prompt injection.
›Adds Human-in-the-Loop (HITL) patterns with improved interrupt handling, including a jump_to helper using end (replacing __end__).
›Adds ToolConfig integration for HITL, allowing interrupts to be conditioned on ToolConfig values.
+3 moreshow less
›Supports PEP 604 (| union) syntax in tool node error handlers.
›Adds stuff and map reduce chains.
›Drops Python 3.9 support for langchain v1.
└──▷ BREAKING ON UPGRADE
!create_react_agent has been renamed to create_agent; code referencing create_react_agent will break.
!Python 3.9 is no longer supported in langchain v1.
!The __end__ value for jump_to has been replaced with end; existing calls using __end__ will break.
$ git clone --branch langchain==1.0.0a8 https://github.com/langchain-ai/langchain.git
# already have the repo? check out this version:$ git checkout langchain==1.0.0a8
└──▷ USE IT
Use the new decorator pattern to register dynamically generated middleware in create_agent for per-request system prompt injection.
python
from langchain import create_agent
@middleware
def dynamic_system_prompt(state, config):
return {"system": f"You are a helpful assistant. Today is {date.today()}."}
agent = create_agent(model, tools, middleware=[dynamic_system_prompt])
›Adds a new decorator pattern for dynamically generated middleware in create_agent.
›Adds create_agent (renamed from create_react_agent) with middleware support, enabling dynamic system prompt middleware and composable agent pipelines.
›Adds dynamic system prompt middleware, allowing runtime-configurable system prompts via middleware nodes.
›Adds improved Human-in-the-Loop (HITL) patterns with simplified conditions and interrupt control gated on ToolConfig values.
›Adds PEP 604 (| union) syntax support in tool node error handlers.
+3 moreshow less
›Adds improvements to Anthropic prompt caching.
›Adds stuff and map_reduce chains.
›Drops Python 3.9 support for the v1 package.
└──▷ BREAKING ON UPGRADE
!create_react_agent has been renamed to create_agent; any code importing or calling create_react_agent from langchain will break.
!Python 3.9 is no longer supported in the langchain v1 package; users on 3.9 must upgrade their runtime.
LlamaIndex v0.14.3 adds ThinkingBlock content support across LLMs, a PaddleOCR reader, Azure PostgreSQL vector store, and Valyu Extractor with Fast mode.
└──▷ GET THIS VERSION
$ git clone --branch v0.14.3 https://github.com/run-llama/llama_index.git
# already have the repo? check out this version:$ git checkout v0.14.3
›New llama-index-readers-paddle-ocr package (PaddleOCR Reader) extracts text from images embedded in PDFs.
›New llama-index-vector-stores-azurepostgresql package adds vector store support for Azure PostgreSQL.
›Adds ThinkingBlock as a supported content block type across llama-index-core, llama-index-llms-anthropic, llama-index-llms-google-genai, llama-index-llms-mistralai, and llama-index-llms-openai.
›Adds Valyu Extractor and Fast mode to llama-index-tools-valyu.
›llama-index-llms-google-genai gains FileAPI support for document uploads, previously missing.
+3 moreshow less
›llama-index-readers-mongodb, llama-index-storage-chat-store-mongo, and llama-index-storage-kvstore-mongodb migrate from Motor to the PyMongo native asynchronous API.
›llama-index-readers-web Firecrawl integration migrates to the Firecrawl v2 SDK.
›llama-index-llms-baseten adds support for the kimik2-0905 model and introduces Dynamic Model APIs validation.
Continue gains CometAPI as a new model provider in the OpenAI adapters package.
└──▷ 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 CometAPI as a supported model provider, enabling routing of completions and chat through the CometAPI backend.
Continue 1.19.0 adds AGENTS.md support, improved MCP connection status, headless error surfacing, and a /share command in VS Code.
└──▷ 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 support for AGENTS.md to define agent-level rules and instructions.
›Headless mode now surfaces errors and exits with code 1 on failure, enabling reliable CI integration.
›Exposes legacy /share command to the VS Code UI for sharing conversations.
›Improves MCP connection status visibility so users can see server connectivity at a glance.
›Adds 'mark unread' capability for chat sessions.
+1 moreshow less
›Cleans up Anthropic messaging handling for improved model communication.
an open source, extensible AI agent that goes beyond code suggestions - install, execute, edit, and test with any LLM
Goose v1.9.0 adds prompt injection detection, auto-compaction, an analyze tool, streaming for Tetrate ARS, and richer recipe management in the UI.
└──▷ GET THIS VERSION
$ git clone --branch v1.9.0 https://github.com/block/goose.git
# already have the repo? check out this version:$ git checkout v1.9.0
›Adds prompt injection detection (pattern-matching) to flag malicious inputs before they reach the model.
›Automatically compacts sessions for streaming providers when context length is exceeded, with an offer to summarize or clear overly long conversations.
›New analyze tool added to the Developer extension for code and file analysis.
›Adds streaming support to the Tetrate Agent Router Service (ARS) provider.
›Supports importing file contents from recipe file input-type parameters.
+8 moreshow less
›Desktop UI now supports uploading JSON recipe files directly.
›Recipe schema validation added to the UI to catch malformed recipes before execution.
›UI recipe activities now support parameter replacement and always appear at the top of chat when a recipe is active.
›Adds Message Metadata for Visibility Control, enabling fine-grained control over which messages are shown.
›I Ching MCP server added to the extension catalog.
›Upgraded agent-client-protocol (ACP) with improved tool experience and nicer tool name formatting.
›Recipe title field added to the import form.
›UI filters invalid, duplicate, and unused parameters from recipes automatically.
└──▷ BREAKING ON UPGRADE
!The --path flag has been removed to avoid session directory validation issues.
Qwen Code v0.1.0-preview adds OpenAI and Qwen OAuth authentication support for the Zed ACP integration.
└──▷ GET THIS VERSION
$ git clone --branch v0.1.0-preview https://github.com/QwenLM/qwen-code.git
# already have the repo? check out this version:$ git checkout v0.1.0-preview
›Adds OpenAI and Qwen OAuth authentication support to the Zed ACP integration.
Qwen Code v0.0.13-nightly.4 adds OpenAI and Qwen OAuth authentication to the Zed ACP integration.
└──▷ GET THIS VERSION
$ git clone --branch v0.0.13-nightly.4 https://github.com/QwenLM/qwen-code.git
# already have the repo? check out this version:$ git checkout v0.0.13-nightly.4
›Adds OpenAI and Qwen OAuth authentication support to the Zed ACP integration.
›Adds whitespace_map setting to control which visible characters render whitespace when show_whitespace is enabled.
›Adds repl.max_columns and repl.max_lines settings to cap REPL output dimensions.
›Adds agent.message_editor_min_lines setting to customize the agent panel message editor minimum height.
›Adds stop_at_newlines option to editor::CutToEndOfLine (ctrl-k) to prevent gobbling newlines.
›Adds pane::SplitAndMoveUp, pane::SplitAndMoveDown, pane::SplitAndMoveLeft, and pane::SplitAndMoveRight actions to create a split without cloning the current buffer.
+18 moreshow less
›Adds a stash picker to pop or drop a specific git stash entry, plus new git stash apply and drop commands.
›Adds built-in support for the ty Python language server (disabled by default) — no extension install required.
›Ships the Ruff native language server built-in for Python (enabled by default) — no extension install required.
›Adds out-of-the-box Tailwind completions in .ts files via TypeScript support.
›Adds API key and API URL configuration for connecting to a remote Ollama provider from agent panel settings.
›Adds Provider Routing control for OpenRouter models via settings.
›Adds dedicated 'helix select' mode in Helix emulation that can be targeted by keybindings.
›Enables commit message amendments in the Git UI without requiring any staged changes in the working copy.
›Adds tooltips to Git branch picker items to distinguish long branch names.
›Agent panel now sends a file outline instead of full file contents when @mentioning large files, reducing token consumption.
›Adds an error message when a Gemini response contains a block_reason.
›Adds scrollbars to the markdown preview and syntax tree view.
›Adds support for icon themes to change folder icons based on directory name.
›ACP now passes proxy settings through to all ACP agents.
›Installing a dev extension now automatically removes the corresponding release extension if installed.
›Allows 0 as a valid value for the excerpt_context_lines setting.
›Improved drag-and-drop in the project panel: files and external paths can be dropped into empty space, and folded directories can be dragged onto other items.
›Python toolchain prioritization improved to prefer local virtual environments.
gemini-cli v0.7.0-preview.0 adds policy engine config, git extension auto-updates, headless custom commands, and global telemetry/debug/proxy flags.
└──▷ GET THIS VERSION
$ git clone --branch v0.7.0-preview.0 https://github.com/google-gemini/gemini-cli.git
# already have the repo? check out this version:$ git checkout v0.7.0-preview.0
└──▷ TRY IT
Restrict which tools Gemini can call, specifying multiple tools in a single comma-separated argument.
$ gemini --allowed-tools read_file,run_shell_command 'Summarize all TODO comments in this repo'
Enable debug output globally alongside a prompt, now that --debug is a global flag.
$ gemini --debug 'What files changed in the last commit?'
›Adds policy engine configuration from existing settings via feat(cli): configure policy engine from existing settings.
›Supports auto-updating git extensions installed from GitHub release URLs.
›Enables custom commands in non-interactive/headless mode.
›Promotes --telemetry, --debug, and --proxy flags to global scope, usable with any subcommand.
›Adds MCP Server name to ToolCallEvent OpenTelemetry logging for better tracing of MCP tool calls.
+6 moreshow less
›Emits a warning when memory usage exceeds 7 GB.
›Adds a simple extensions release flow to support publishing extensions.
›Supports rendering function calls and responses when sharing a chat session to Markdown.
›Correctly supports comma-separated values for --allowed-tools.
›Hardens the VS Code IDE companion server with CORS and host validation.
›Improves parsing of GitHub extension source URIs.
└──▷ BREAKING ON UPGRADE
!The shell setting usePty is renamed to enableInteractiveShell; existing configs using usePty will stop working.
Get up and running with Kimi-K2.6, GLM-5.2, MiniMax, DeepSeek, gpt-oss, Qwen, Gemma and other models.
Ollama v0.12.2 adds a web search API to ground model responses in live data and expands new-engine model support.
└──▷ GET THIS VERSION
$ git clone --branch v0.12.2 https://github.com/ollama/ollama.git
# already have the repo? check out this version:$ git checkout v0.12.2
›New web search API lets models retrieve live web results to reduce hallucinations and improve accuracy, with a free tier for individuals and higher rate limits via Ollama Cloud.
›Qwen3 architecture models, including MoE variants, now run on Ollama's new engine.
›Supports multi-regex pretokenizers in Ollama's new engine.
›Ollama's new engine can load tensors by matching a prefix or suffix.
Phoenix playground gains custom HTTP header support for LLM proxy and gateway scenarios.
└──▷ GET THIS VERSION
$ git clone --branch arize-phoenix-v11.37.0 https://github.com/Arize-ai/phoenix.git
# already have the repo? check out this version:$ git checkout arize-phoenix-v11.37.0
›Adds custom HTTP headers for playground requests, enabling auth tokens and routing headers to be sent when calling LLM endpoints through proxies or gateways.
›Adds a new checkbox UI control (details unspecified in release notes).
arize-phoenix-evals 2.1.0 adds Azure provider support for LLM-based evaluations.
└──▷ GET THIS VERSION
$ git clone --branch arize-phoenix-evals-v2.1.0 https://github.com/Arize-ai/phoenix.git
# already have the repo? check out this version:$ git checkout arize-phoenix-evals-v2.1.0
›Adds Azure provider support, enabling evals to run against Azure-hosted LLM endpoints.
LanceDB python-v0.25.2-beta.0 adds use_index parameter to merge insert and namespace-backed database support.
└──▷ GET THIS VERSION
$ git clone --branch python-v0.25.2-beta.0 https://github.com/lancedb/lancedb.git
# already have the repo? check out this version:$ git checkout python-v0.25.2-beta.0
›Adds use_index parameter to merge insert operations, letting callers control whether an index is used during the merge step.
›Adds namespace-backed database support in the Rust backend, enabling namespace-scoped database connections.
LanceDB v0.22.2-beta.0 adds use_index parameter to merge-insert and namespace-backed database support in Rust.
└──▷ GET THIS VERSION
$ git clone --branch v0.22.2-beta.0 https://github.com/lancedb/lancedb.git
# already have the repo? check out this version:$ git checkout v0.22.2-beta.0
›Adds use_index parameter to merge insert operations, allowing callers to control whether an index is used during the merge.
›Adds namespace-backed database support in the Rust client, enabling namespace-scoped database connections.
THE AI TOOLCHAIN