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 v2.6.5 adds Gemini multimodal file search, Gmail/Calendar context providers, Mongo scheduler support, and new workflow error handling.
└──▷ GET THIS VERSION
$ git clone --branch v2.6.5 https://github.com/agno-agi/agno.git
# already have the repo? check out this version:$ git checkout v2.6.5
└──▷ USE IT
Restrict an agent's LLMsTxt fetching to only trusted domains to prevent SSRF-style abuse in agentic pipelines.
python
from agno.tools.llms_txt import LLMsTxtTools
tools = LLMsTxtTools(allowed_hosts=["docs.example.com", "api.example.com"])
Enable file download and upload capabilities in a Slack-connected agent for workflows that need to handle attachments.
python
from agno.context.slack import SlackContextProvider
provider = SlackContextProvider(enable_media_tools=True)
›Adds allowed_hosts parameter to LLMsTxtTools so agents only fetch from explicitly trusted hosts.
›Adds enable_media_tools flag (default: False) to SlackContextProvider to control file download/upload; when enabled, exposes download_file in read tools and upload_file in write tools.
›Adds on_error handling to the Condition workflow step, giving control over error propagation when sub-steps fail.
›Adds GmailContextProvider and CalendarContextProvider, following the same pattern as existing GDriveContextProvider, SlackContextProvider, and DatabaseContextProvider.
›Extends GDriveContextProvider to support OAuth authentication in addition to service account auth.
+2 moreshow less
›Adds scheduler support for MongoDb and AsyncMongoDb backends, enabling agents, teams, and workflows to run on a cron schedule in AgentOS.
›Adds multimodal support in the Gemini File Search API (requires google-genai>=1.75.0), enabling image and other media types alongside text in file search workflows.
LangChain 1.3.0a2 adds stream_events v3, a respond decision in HITL middleware, dynamic tool registration, and ToolCallRequest exports.
└──▷ GET THIS VERSION
$ git clone --branch langchain==1.3.0a2 https://github.com/langchain-ai/langchain.git
# already have the repo? check out this version:$ git checkout langchain==1.3.0a2
└──▷ USE IT
Stream agent events using the new v3 protocol for finer-grained content-block streaming in production agent pipelines.
python
async for event in agent.astream_events(input, version='v3'):
print(event)
Use the respond decision in HITL middleware to let middleware return a final answer without invoking further tools.
python
from langchain.agents.middleware import respond
class MyHITLMiddleware:
def on_tool_call(self, request):
if needs_human_approval(request):
return respond('Action blocked pending human review.')
›Wires stream_events(version='v3') into create_agent, enabling the new v3 streaming protocol end-to-end in agent flows.
›Adds respond decision to the human-in-the-loop (HITL) middleware, letting middleware short-circuit agent execution and return a response directly.
›Adds ToolCallRequest to middleware exports, giving middleware code direct access to the structured tool-call request object.
›Supports dynamic tool registration via middleware, allowing tools to be added or removed at runtime during an agent run.
›Adds state field to _ModelRequestOverrides, enabling per-request state overrides when calling the model through middleware.
+7 moreshow less
›Adds threading context propagation through create_agent flows and middleware.
›Adds ls_agent_type tag on create_agent calls for improved LangSmith tracing and agent-type classification.
›Adds LangSmith integration metadata to create_agent and init_chat_model for richer observability.
›Supports state updates from wrap_model_call with command(s), enabling graph state mutations from within model-call wrappers.
›Adds tracing for wrap_model_call and tool calls.
›Adds content-block-centric streaming (version='v2') protocol to langchain-core.
›Adds langchain-openrouter provider package for routing requests through OpenRouter.
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 v2.1.132 adds session ID env var in subprocesses and an opt-out for the fullscreen alternate-screen renderer.
└──▷ GET THIS VERSION
$ git clone --branch v2.1.132 https://github.com/anthropics/claude-code.git
# already have the repo? check out this version:$ git checkout v2.1.132
└──▷ TRY IT
Keep Claude Code output in your terminal's native scrollback instead of the fullscreen renderer — useful in tmux, screen, or any workflow where you rely on scroll history.
$ CLAUDE_CODE_DISABLE_ALTERNATE_SCREEN=1 claude
Reference the session ID inside a Bash tool script to tag logs or correlate subprocess events back to the originating Claude Code session.
$ echo "Running inside Claude Code session: $CLAUDE_CODE_SESSION_ID" >> /tmp/session-audit.log
›Adds CLAUDE_CODE_SESSION_ID environment variable to Bash tool subprocess environments, matching the session_id passed to hooks — enabling scripts and hooks to correlate subprocess activity back to the parent session.
›Adds CLAUDE_CODE_DISABLE_ALTERNATE_SCREEN=1 env var to opt out of the fullscreen alternate-screen renderer, keeping conversation output in the terminal's native scrollback.
›Adds a 'Pasting…' footer hint while a Ctrl+V image paste is being read from the clipboard.
Claude Code v2.1.129 adds URL-based plugin loading, auto-update on Homebrew/WinGet, and cross-project history search by default.
└──▷ GET THIS VERSION
$ git clone --branch v2.1.129 https://github.com/anthropics/claude-code.git
# already have the repo? check out this version:$ git checkout v2.1.129
└──▷ TRY IT
Load a community or internal plugin directly from a URL without installing it permanently.
$ claude --plugin-url https://example.com/my-plugin.zip
Enable background auto-updates on a Homebrew install so Claude Code silently upgrades and prompts to restart.
$ export CLAUDE_CODE_PACKAGE_MANAGER_AUTO_UPDATE=1
claude
Force synchronized output when running Claude Code inside Emacs eat or another terminal where auto-detection fails.
$ CLAUDE_CODE_FORCE_SYNC_OUTPUT=1 claude
›Adds --plugin-url <url> flag to fetch and load a plugin .zip archive from a URL for the current session.
›Adds CLAUDE_CODE_FORCE_SYNC_OUTPUT=1 env var to force synchronized output on terminals where auto-detection fails (e.g. Emacs eat).
›Adds CLAUDE_CODE_PACKAGE_MANAGER_AUTO_UPDATE env var: when set on Homebrew or WinGet installs, Claude Code upgrades itself in the background and prompts to restart.
›Adds CLAUDE_CODE_ENABLE_GATEWAY_MODEL_DISCOVERY=1 env var to opt in to Gateway /v1/models discovery for the /model picker (now opt-in instead of automatic).
›Ctrl+R history picker now searches all prompts across all projects by default; press Ctrl+S to narrow to the current project or session.
+4 moreshow less
›skillOverrides setting is now functional: off hides a skill from the model and /, user-invocable-only hides from the model only, name-only collapses the description.
›The claude_code.pull_request.count OTel metric now counts PRs/MRs created via MCP tools in addition to shell commands.
›Policy refusal error messages now include the API Request ID to aid support debugging.
›Plugin manifests: themes and monitors should now be declared under "experimental": { ... }; claude plugin validate warns on top-level declarations.
└──▷ BREAKING ON UPGRADE
!Gateway /v1/models model discovery (automatic in v2.1.126–2.1.128) is now opt-in via CLAUDE_CODE_ENABLE_GATEWAY_MODEL_DISCOVERY=1; setups relying on automatic gateway model discovery will no longer see those models without setting this variable.
Zed v1.1.5 adds LSP code lens, git graph view, split diff, panel layout switcher, Helix amp jump, and a line_ending setting with .editorconfig support.
└──▷ GET THIS VERSION
$ git clone --branch v1.1.5 https://github.com/zed-industries/zed.git
# already have the repo? check out this version:$ git checkout v1.1.5
└──▷ USE IT
Enable LSP code lens in Zed to see inline references and run/test actions directly in the editor.
json
{
"code_lens": "on"
}
Force polling filesystem watching in WSL where inotify events are unavailable, so Zed picks up file changes on network or FUSE mounts.
📍ZED_FILE_WATCHER_MODE=poll zed .
›Adds LSP code lens support (disabled by default); enable with "code_lens": "on" in settings.
›Adds line_ending setting to control how line endings are handled for new files and normalized on save.
›Adds support for .editorconfigend_of_line property to enforce line endings per project.
›Adds allow_data_collection key under edit_predictions in settings.json to explicitly control Edit Predictions data collection.
›Adds go_to_definition_scroll_strategy setting to configure how go-to-definition scrolls the editor (default now centers the target).
+23 moreshow less
›Adds helix.jump_label_accent setting to control the color of amp jump labels in Helix mode.
›Adds markdown_preview_font_family and markdown_preview_theme settings to customize markdown preview independently from the editor.
›Adds credentials_url setting for remote development, used as the key for storing credentials in Keychain Access, falling back to server_url.
›Adds ZED_FILE_WATCHER_MODE=poll environment variable to force polling filesystem watching on Linux/WSL for filesystems without inotify support (e.g. /mnt/c, network filesystems, FUSE mounts); polling is also chosen automatically when needed.
›Adds audible BEL to the terminal; enable by setting terminal.bell to "system".
›Adds a Panel Layout switcher in the user menu with classic and agentic presets for quick layout swaps.
›Adds Helix mode 'amp jump' navigation (g w) for two-character label-based word jumping with forward/backward prioritization.
›Replaces the file history view with a git graph view featuring lazy loading, search, resizable columns, and a commit information panel.
›Adds side-by-side split diff view in both the agent review diff and file diff panels.
›Adds 'Always allow' tool approval in the agent panel that propagates to other pending calls to the same tool.
›Adds a git: copy branch name action.
›Adds a dedicated empty state in the Git Panel for unsafe repositories with a 'Trust Directory' button that adds the repo to safe.directory.
›Adds a warning when a git commit message title exceeds a configurable character limit.
›Adds support for GFM alert callouts (> [!NOTE], > [!TIP], > [!IMPORTANT], > [!WARNING], > [!CAUTION]) in markdown preview, rendered with colored left borders, icons, and bold labels.
›Adds 'Add to .gitignore' option to the project panel context menu for files in git repositories.
›Adds OpenCode Go provider support, including configurable reasoning effort levels.
›Adds support for interleaved_reasoning in OpenCode.
›Adds new OpenCode models: GPT 5.5, DeepSeek V4, MiMo V2.5, Ling 2.6, and Hy3.
›Adds Copilot edit prediction support in temporary files.
›Adds a button to stop commit message generation in the git panel.
›Adds vim mode navigation to the git graph view.
›Makes table-cell markdown checkboxes clickable in markdown preview.
›Project directories now preserve the order they were added rather than sorting lexicographically.
└──▷ BREAKING ON UPGRADE
!Go-to-definition now opens its target centered in the editor by default; the previous scroll behavior must be restored via the new go_to_definition_scroll_strategy setting.
llama.cpp b9045 adds multimodal speech support for IBM Granite 4.0 1B Speech via the mtmd layer.
└──▷ GET THIS VERSION
$ git clone --branch b9045 https://github.com/ggml-org/llama.cpp.git
# already have the repo? check out this version:$ git checkout b9045
›Adds ibm-granite/granite-4.0-1b-speech model support in mtmd, enabling Conformer-encoder speech-to-text with QFormer projector, 80-bin log-mel spectrogram, and 2x frame stacking (80→160 mel dims).
›GGUF converter (convert) now handles batch-norm folding, fused K/V split, and Conv1d weight reshaping for Granite Speech models at export time, and registers GraniteModel for GraniteSpeechForConditionalGeneration.
llama.cpp b9038 improves OpenCL memory estimation for --fit using CL_DEVICE_GLOBAL_MEM_SIZE.
└──▷ GET THIS VERSION
$ git clone --branch b9038 https://github.com/ggml-org/llama.cpp.git
# already have the repo? check out this version:$ git checkout b9038
›Uses CL_DEVICE_GLOBAL_MEM_SIZE to report estimated OpenCL device memory when using --fit, enabling more accurate model-to-device fitting decisions on OpenCL backends.
THE AI TOOLCHAIN