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.
Continue v1.2.17 adds AI SDK provider integration, background bash execution, hooks system, turn-level prompt caching, and z.ai support.
└──▷ GET THIS VERSION
$ git clone --branch v1.2.17-vscode https://github.com/continuedev/continue.git
# already have the repo? check out this version:$ git checkout v1.2.17-vscode
›Adds environment variable to enable the AI SDK provider (feat: use env variable to enable ai sdk).
›Adds ai-sdk provider integration, with support for xai and deepseek backends.
›Adds a hooks system for CLI event interception, enabling programmatic interception of agent events.
›Enables bash tool execution in the background, with background job permission checks allowed by default.
›Adds turn-level prompt caching and conversation message caching to AnthropicApi, reducing token costs on repeated context.
+6 moreshow less
›Adds /rename as an alias for the /title CLI command.
›Moves agent configuration directory from .continue/agents to .continue/checks.
›Adds 5 new agent checks derived from codebase history.
›Adds support for z.ai as a model provider, alongside Claude Sonnet and Opus 4.6 model updates.
›Adds pre-install suggestions CLI in devbox blueprints.
›Adds quiz answers subcommand to the CLI.
└──▷ BREAKING ON UPGRADE
!Agent configuration directory is moved from .continue/agents to .continue/checks; existing agent configs must be relocated.
Crush v0.49.0 adds transparent-mode toggle, AWS Bedrock API key pasting, and configurable Bash tool auto-background timing.
└──▷ GET THIS VERSION
$ git clone --branch v0.49.0 https://github.com/charmbracelet/crush.git
# already have the repo? check out this version:$ git checkout v0.49.0
›Adds a command-palette (ctrl+p) toggle to enable/disable background color ('transparent mode'), persisted across sessions.
›Supports pasting AWS Bedrock API keys directly in the provider input, alongside the existing environment-variable method.
›Adds auto_background_after parameter to the Bash tool, letting the model specify how many seconds before a command is moved to a background job (default: 60s).
Copilot CLI v1.0.5 adds /pr for pull-request automation, /extensions management, richer file mentions, and a preCompact hook.
└──▷ GET THIS VERSION
$ git clone --branch v1.0.5 https://github.com/github/copilot-cli.git
# already have the repo? check out this version:$ git checkout v1.0.5
└──▷ TRY IT
Kick off automated PR creation and CI-fix loop from inside a Copilot CLI session.
$ /pr
Browse recent release notes or summarize a range of versions without leaving the session.
$ /changelog last 5
/changelog since 1.0.0
/changelog summarize
›Adds /pr command to create and view pull requests, automatically fix CI failures, address review feedback, and resolve merge conflicts.
›Adds /extensions command to view, enable, and disable CLI extensions from within the session.
›Adds /version command to display the CLI version and check for updates without leaving the session.
›Extends /changelog with last <N>, since <version>, and summarize subcommands to browse and summarize multiple release notes at once.
›Adds preCompact hook to run commands before context compaction starts.
+7 moreshow less
›Expands @ file mentions to support paths outside the project: absolute paths (@/usr/...), home directory (@~/...), and relative parent paths (@../...).
›Enables multi-turn conversations with background agents via the write_agent tool for sending follow-up messages.
›Syntax highlighting in /diff now covers 17 programming languages.
›Toggleing experimental mode with /experimental on|off automatically restarts the CLI to apply changes immediately.
›Shows a clear error when a classic Personal Access Token (ghp_) is set in environment variables instead of silently exiting.
›Surfaces the API request ID in the timeline when errors occur after retries are exhausted.
›Hooks config files that omit the version field are now accepted by the CLI.
langchain-mistralai 1.1.2 adds text_inputs and text_outputs fields to model profiles.
└──▷ GET THIS VERSION
$ git clone --branch langchain-mistralai==1.1.2 https://github.com/langchain-ai/langchain.git
# already have the repo? check out this version:$ git checkout langchain-mistralai==1.1.2
›Adds text_inputs and text_outputs fields to model profiles, expanding the metadata available for Mistral model configuration.
OpenClaw v2026.3.12 adds a revamped gateway dashboard, GPT-5.4/Claude fast-mode toggles, Slack Block Kit support, and a new sessions_yield orchestration primitive.
└──▷ GET THIS VERSION
$ git clone --branch v2026.3.12 https://github.com/openclaw/openclaw.git
# already have the repo? check out this version:$ git checkout v2026.3.12
›New modular gateway dashboard (dashboard-v2) with overview, chat, config, agent, and session views; command palette; mobile bottom tabs; and chat tools including slash commands, search, export, and pinned messages.
›Adds configurable session-level fast-mode toggle (/fast) across TUI, Control UI, and ACP for OpenAI/GPT-5.4, with per-model config defaults and Codex request shaping.
›Maps /fast toggle and params.fastMode to Anthropic API service_tier requests for Claude fast mode, with live tier verification for both Anthropic and OpenAI.
›Moves Ollama, vLLM, and SGLang onto a provider-plugin architecture with provider-owned onboarding, discovery, model-picker setup, and post-selection hooks.
›Adds a Kubernetes install path with raw manifests, Kind setup, and deployment docs.
+2 moreshow less
›Adds sessions_yield so orchestrators can end the current turn immediately, skip queued tool work, and carry a hidden follow-up payload into the next session turn.
›Supports channelData.slack.blocks in the shared reply delivery path, enabling agents to send Slack Block Kit messages through standard outbound delivery.
Semantic Kernel Python 1.41.0 adds support for the new OpenAI text-to-image model.
└──▷ GET THIS VERSION
$ git clone --branch python-1.41.0 https://github.com/microsoft/semantic-kernel.git
# already have the repo? check out this version:$ git checkout python-1.41.0
›Adds support for the new OpenAI text-to-image model in the Python SDK.
test-backend-ops gains file-based test loading and a graph operator extraction tool via llama_graph_reserve
└──▷ GET THIS VERSION
$ git clone --branch b8320 https://github.com/ggml-org/llama.cpp.git
# already have the repo? check out this version:$ git checkout b8320
›Adds llama_graph_reserve (moved to new llama-ext header) as the public API for reserving graph computations, replacing previous internal API calls.
›Adds file-based test loading to test-backend-ops, allowing backend operation tests to be defined and loaded from a text file instead of being hardcoded.
›Adds the export-graph-ops tool (located in tests/) to extract model graph operators into a file for use with the new file-based test runner.
›Adds per-operator error thresholds to test-backend-ops for more precise correctness validation across different backend operations.
›Adds nb parameter support in test-backend-ops for non-contiguous input tensors.
llama.cpp b8317 adds Vulkan GATED_DELTA_NET op support for AMD GPUs via fused compute shader.
└──▷ GET THIS VERSION
$ git clone --branch b8317 https://github.com/ggml-org/llama.cpp.git
# already have the repo? check out this version:$ git checkout b8317
›Adds GATED_DELTA_NET op to the Vulkan backend as a fused compute shader, supporting scalar gate, KDA vector gate, GQA broadcast, multi-token sequences, and permuted (non-contiguous) q/k inputs — enabling Gated Delta Network model inference on Vulkan GPUs.
›Vulkan GATED_DELTA_NET shader uses specialization constants to select head size (32, 64, or 128) and KDA mode at pipeline creation time, with a pipeline array layout of [3][2].
›Vulkan GATED_DELTA_NET KDA path caches exp(g) in shared memory and uses vec4 dot products on all inner loops, yielding +5.4% throughput on the KDA path on AMD Radeon 890M (RADV GFX1150).
llama.cpp b8303 adds support for the Phi4ForCausalLMV multimodal model architecture.
└──▷ GET THIS VERSION
$ git clone --branch b8303 https://github.com/ggml-org/llama.cpp.git
# already have the repo? check out this version:$ git checkout b8303
›Adds support for the Phi4ForCausalLMV model architecture, enabling Phi-4 vision inference with corrected SigLIP2 patch-kernel layout and NaFlex resize behavior.
llama.cpp b8299 adds GPU-accelerated DeltaNet inference via fused Metal and CUDA kernels for GDN, boosting Qwen3.5 throughput by up to 25%.
└──▷ GET THIS VERSION
$ git clone --branch b8299 https://github.com/ggml-org/llama.cpp.git
# already have the repo? check out this version:$ git checkout b8299
›Adds fused Metal kernel for GGML_OP_GATED_DELTA_NET on Apple Silicon, enabling GPU-accelerated inference for DeltaNet-based models (Qwen3.5, etc.) with both GDA (scalar gate) and KDA (per-row gate) modes; supports head_size 64 and 128, falling back to CPU for head_size 32 or non-contiguous tensors — yields +25% token generation throughput (170 → 213 t/s) for Qwen3.5-0.8B Q4_K_M on M4 Max.
›Enables chunked fused GDN path in the llama inference engine, extending DeltaNet recurrence support to prefill in addition to token generation.
llama.cpp b8297 adds NVFP4 quantization type with ARM NEON optimization delivering up to 3.8x throughput gains.
└──▷ GET THIS VERSION
$ git clone --branch b8297 https://github.com/ggml-org/llama.cpp.git
# already have the repo? check out this version:$ git checkout b8297
›Adds NVFP4 as a new quantization type option in llama-quantize, with CPU fallback dot product, quantization, dequantization, and conversion support across all platforms (ARM, x86, PowerPC, s390x).
›Adds optimized ggml_vec_dot_nvfp4_q8_0 ARM NEON implementation using vqtbl1q_s8 LUT lookup, vpaddq_s32 reduction, and vfmaq_f32 accumulators — achieving ~31.0 t/s (tg128) versus 8.1 t/s generic, a 3.8x speedup at 77% of Q4_1 speed.
›Stores NVFP4 scales as native UE4M3 format, preserving original model bits when converting from HuggingFace via convert_hf_to_gguf.py.
›NVFP4 CPU-only backend uses a 64-block super-block layout; PPL benchmarks at 15.25 versus F16 at 14.33 at 4.70 BPW, comparable to Q4_1 (15.81) at lower bit-width.
Phoenix 13.15.0 adds phoenix db migrate, a traces REST endpoint, and new span filters.
└──▷ GET THIS VERSION
$ git clone --branch arize-phoenix-v13.15.0 https://github.com/Arize-ai/phoenix.git
# already have the repo? check out this version:$ git checkout arize-phoenix-v13.15.0
└──▷ TRY IT
Run database migrations independently of the Phoenix server, useful in CI or containerised deploy pipelines.
$ phoenix db migrate
›Adds phoenix db migrate subcommand for running database migrations standalone, decoupled from server startup.
›Adds GET /v1/projects/{project_identifier}/traces REST endpoint for retrieving traces per project.
›Adds name, span_kind, and status_code filter parameters to the REST spans API.
Langfuse v3.158.0 adds eval config blocking, unresolved prompt fetches in the v2 API, a blob storage status endpoint, and fulltext search in the playground.
└──▷ GET THIS VERSION
$ git clone --branch v3.158.0 https://github.com/langfuse/langfuse.git
# already have the repo? check out this version:$ git checkout v3.158.0
›Adds unresolved prompt fetches to the v2 prompts API, exposing fetch state directly in API responses.
›Adds a blob storage integration status endpoint with error tracking for diagnosing storage integration health.
›Adds eval config blocking to prevent misconfigured evaluators from running.
›Enables fulltext search across message windows in the playground and prompt editor.
›Allows project members (not just admins) to edit LLM tools in the UI.
+2 moreshow less
›Adds media and blob batch cleaner that runs on project deletion to clean up associated storage objects.
›Introduces an intro dialog for the Fast (Preview) dashboard toggle, now promoted from beta to preview.
Composio CLI gains --no-wait and --key flags on the login command to support headless agent and auth flows.
└──▷ GET THIS VERSION
$ git clone --branch @composio/[email protected] https://github.com/ComposioHQ/composio.git
# already have the repo? check out this version:$ git checkout @composio/[email protected]
└──▷ TRY IT
Kick off a headless login in a CI/agent pipeline — print the auth URL and session info without blocking, then complete it later with a known session key.
$ composio login --no-wait--key <session-key>
Obtain a login URL non-interactively so an orchestration script can present it elsewhere, then exit without waiting.
$ composio login --no-wait
›Adds --no-wait flag to the login command to print the URL and session info then exit immediately, enabling non-interactive agent flows.
›Adds --key flag to the login command to complete login with a session key, polling until the session is linked unless --no-wait is also passed.
Composio CLI v0.2.2 adds interactive org/project picker, --no-wait for composio link, and redacts API keys from whoami
└──▷ GET THIS VERSION
$ git clone --branch @composio/[email protected] https://github.com/ComposioHQ/composio.git
# already have the repo? check out this version:$ git checkout @composio/[email protected]
└──▷ TRY IT
Emit the link URL and exit immediately in CI pipelines instead of waiting for the link to complete.
$ composio link --no-wait
Run a fully non-interactive login in automation by skipping the org/project picker and completing auth via browser flow.
$ composio login -y
›Adds --no-wait flag to composio link — prints URL/JSON output and exits immediately without blocking on completion.
›Adds -y flag to composio login to skip the new interactive org/project picker and proceed non-interactively via browser flow.
›Adds composio install command for shell integration.
›Adds interactive org/project picker after composio login to guide users through org and project selection.
›Adds background upgrade-available hint to the CLI to surface when a newer version is available.
+1 moreshow less
›composio whoami no longer exposes API keys in its output.
└──▷ BREAKING ON UPGRADE
!The --api-key, --org-id, and --project-id flags have been removed from composio login and composio init; non-interactive login via these flags is no longer supported — use the browser flow with -y instead.
THE AI TOOLCHAIN