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.
AutoGPT Platform adds Slack and Telegram adapters, first-class org/workspace support, and a Copilot composer with guided creation flows.
└──▷ GET THIS VERSION
$ git clone --branch autogpt-platform-beta-v0.6.67 https://github.com/Significant-Gravitas/AutoGPT.git
# already have the repo? check out this version:$ git checkout autogpt-platform-beta-v0.6.67
›Adds Slack adapter supporting webhook and Events API integration, with multi-workspace install via OAuth and visibility on the Bots settings and admin analytics pages.
›Adds Telegram adapter as a third platform on the chat bus, with live draft previews that stream replies in real time.
›Introduces first-class org/workspace support with new schema, auth, APIs, migration, and frontend.
›Adds Copilot composer and menu with Skills, Scheduled, and Integrations modals and guided creation flows.
›Splits adapter base into Socket/Webhook variants and extracts shared adapter helpers (attachments, mentions, history, chunking) to support additional platform integrations.
+2 moreshow less
›Adds LaunchDarkly per-platform visibility control for Settings → Bots.
›Removes built-in Revid credential — users must now provide their own key.
└──▷ BREAKING ON UPGRADE
!The built-in Revid credential has been removed; existing setups using it will break until a user-provided key is supplied.
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 v2.1.211 adds --forward-subagent-text flag and CLAUDE_CODE_FORWARD_SUBAGENT_TEXT env var for subagent stream output, plus several workflow improvements.
└──▷ GET THIS VERSION
$ git clone --branch v2.1.211 https://github.com/anthropics/claude-code.git
# already have the repo? check out this version:$ git checkout v2.1.211
└──▷ TRY IT
Set token budget with scientific notation in an environment where the variable must be typed or scripted compactly.
$ export CLAUDE_MAX_TOKENS=1e6
claude -p 'Summarize this large codebase'
›Adds --forward-subagent-text flag and CLAUDE_CODE_FORWARD_SUBAGENT_TEXT environment variable to include subagent text and thinking in stream-json output.
›Integer environment variables (timeouts, token budgets, retry counts) now accept scientific notation and digit-separator spellings like 1e6 and 64_000.
›Changes 'always allow' permission rules to save at the repository root, so approvals granted in a git worktree persist across sessions and worktrees.
›Changes /usage-credits to ask for confirmation before sending a request to organization admins.
›The save_to_disk option on Claude in Chrome screenshot actions now writes the image to disk and returns the path (previously did nothing).
+2 moreshow less
›Changes Vim mode s and S (substitute char/line) to work in NORMAL mode, matching standard vim behavior.
›Improves background agent result reporting — Claude now reports the status of still-running agents and waits for real completion instead of fabricating results.
Zed v1.11.3 adds dedicated Staged/Unstaged Changes multibuffers, configurable Git Graph columns, Markdown commit messages, and a project symbols preview pane.
└──▷ GET THIS VERSION
$ git clone --branch v1.11.3 https://github.com/zed-industries/zed.git
# already have the repo? check out this version:$ git checkout v1.11.3
└──▷ HOW TO FIND IT
Review only staged changes as a multibuffer to stage, unstage, or restore individual hunks before committing.
📍git: view staged changes
Forward mouse clicks to vim or another mouse-reporting terminal app instead of opening links on Cmd/Ctrl-click.
json
{
"terminal.open_links_in_mouse_mode": false
}
›Adds git: view staged changes and git: view unstaged changes actions that open dedicated Staged Changes and Unstaged Changes multibuffers for reviewing, staging, unstaging, or restoring individual hunks.
›Adds /helpful and /not-helpful slash commands (shown as 'Positive Feedback' and 'Negative Feedback') in the agent panel message editor for submitting thread feedback.
›Adds protocol validation for OpenCode settings, reasoning effort levels, and parallel tool calls enabled by default for OpenCode, along with new OpenCode Zen models (Fable 5, Sonnet 5, GLM 5.2, Kimi K2.7 Code, Minimax M3).
›Adds terminal.open_links_in_mouse_mode setting to control whether Cmd/Ctrl-click opens links or forwards clicks to the application when mouse reporting is enabled (e.g. vim, opencode, claude).
›Adds boolean toggle support in ACP session configuration for agents that use them.
+15 moreshow less
›Adds 'Agent Panel' and 'Git Panel' items to the 'View' menu.
›Adds Claude Fable 5 and Sonnet 5 to the Amazon Bedrock provider.
›Adds native support for AWS Bedrock's Mantle endpoint, enabling GPT-5.5, GPT-5.4, and Grok 4.3 through the Amazon Bedrock provider.
›Adds a preview pane to the project symbols picker.
›Adds Git Graph column visibility toggle via right-clicking the table header to show or hide columns.
›Renders full commit messages as Markdown in the Git Graph details panel.
›Adds tag labels to the Git Panel commit history.
›Adds diff stat numbers to the uncommitted changes view.
›Adds support for searching inside agent terminal threads with cmd-f (macOS) or ctrl-f (Linux/Windows).
›Adds a copy button in the agent panel to copy the last agent response, and moves 'Open Thread as Markdown' into the agent panel menu.
›Adds the ability to collapse per-file match groups in Text Finder by clicking the chevron on file headers.
›Adds support for middle-clicking a file in the project panel to open it in a permanent tab instead of a preview tab.
›Adds a zed: get merch action that opens the Zed merch store.
›Adds a download progress indicator to the update button.
llama.cpp b10032 adds CUDA implementation of GGML_OP_LIGHTNING_INDEXER with generic vector and WMMA kernels (Turing+).
└──▷ GET THIS VERSION
$ git clone --branch b10032 https://github.com/ggml-org/llama.cpp.git
# already have the repo? check out this version:$ git checkout b10032
›Adds CUDA implementation of GGML_OP_LIGHTNING_INDEXER with a generic vector kernel and a WMMA (Tensor Core) kernel, requiring Turing-class GPUs or newer.
llama.cpp b10016 adds SYCL Flash Attention via oneDNN XMX engine on Intel Battlemage (Xe2), delivering up to 4.26× prefill speedup at 80k context.
└──▷ GET THIS VERSION
$ git clone --branch b10016 https://github.com/ggml-org/llama.cpp.git
# already have the repo? check out this version:$ git checkout b10016
›Adds F16 Flash Attention for SYCL backends using the oneDNN graph API's XMX engine, scoped to Intel Battlemage (Xe2/BMG) GPUs; other architectures fall back to the existing FA kernel.
›Multi-GPU synchronization for the oneDNN FA path via stream wait_and_throw(), enabling validated multi-device inference on Battlemage.
›Delivers up to 4.26× prefill throughput increase at 80k context (pp512: ~32% gain) for Q8_0 models on supported hardware, unlocking long-context use cases that were previously throughput-constrained.
THE AI TOOLCHAIN