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.
langchain-anthropic 1.3.4 adds a ChatAnthropicBedrock wrapper and User-Agent header support for Anthropic API calls.
└──▷ GET THIS VERSION
$ git clone --branch langchain-anthropic==1.3.4 https://github.com/langchain-ai/langchain.git
# already have the repo? check out this version:$ git checkout langchain-anthropic==1.3.4
›Adds User-Agent header to all Anthropic API calls for improved request attribution and observability.
›Adds first-class kilocode provider with auth, onboarding, implicit detection, model defaults, and transcript/cache-ttl handling; default model is kilocode/anthropic/claude-opus-4.6.
›Adds provider: "kimi" (Moonshot) support in the web_search tool, with key/config schema wiring and a two-step $web_search flow that includes citation extraction.
›Adds native Moonshot video provider and includes Moonshot in auto video key detection.
›Accepts Claude shorthand model refs (vercel-ai-gateway/claude-*) for the Vercel AI Gateway by normalizing to canonical Anthropic-routed model IDs.
›Adds optional gateway.http.securityHeaders.strictTransportSecurity config to emit Strict-Transport-Security headers for direct HTTPS deployments.
+3 moreshow less
›Adds openclaw sessions cleanup command with per-agent store targeting, disk-budget controls (session.maintenance.maxDiskBytes / highWaterBytes), and safer transcript/archive cleanup.
›Supports per-agent params overrides merged on top of model defaults (including cacheRetention) for independent cache-behavior tuning per agent.
›Caches bootstrap file snapshots (AGENTS.md/MEMORY.md) per session key and clears them on session reset/delete, reducing prompt-cache invalidations.
└──▷ BREAKING ON UPGRADE
!The browser SSRF policy now defaults to trusted-network mode: browser.ssrfPolicy.dangerouslyAllowPrivateNetwork defaults to true when unset, and the canonical config key is browser.ssrfPolicy.dangerouslyAllowPrivateNetwork instead of the former browser.ssrfPolicy.allowPrivateNetwork. Run openclaw doctor --fix to migrate the legacy key automatically.
HolmesGPT 0.20.0 adds SQL database toolsets, five MCP server integrations, Inspektor Gadget, KubeVela, and a generic HTTP connector toolset.
└──▷ GET THIS VERSION
$ git clone --branch 0.20.0 https://github.com/HolmesGPT/holmesgpt.git
# already have the repo? check out this version:$ git checkout 0.20.0
›Adds SQL Database toolsets covering PostgreSQL, ClickHouse, MariaDB, MySQL, SQLServer, and SQLite, enabling AI-driven investigation of database state.
›Adds a generic HTTP connector toolset for querying arbitrary HTTP endpoints as investigation sources.
›Adds Inspektor Gadget toolset for kernel-level eBPF-based diagnostics within HolmesGPT investigations.
›Adds KubeVela built-in toolsets for investigating OAM application workloads.
›Adds Grafana MCP server integration, enabling HolmesGPT to query Grafana via the Model Context Protocol.
+8 moreshow less
›Adds Sentry MCP server integration for pulling error and event context from Sentry during investigations.
›Adds Prefect MCP server integration for workflow observability via the Model Context Protocol.
›Adds Confluence MCP server integration for surfacing runbook and knowledge-base content during investigations.
›Adds Kubernetes Remediation MCP integration, allowing HolmesGPT to execute remediation actions against Kubernetes clusters.
›Adds filesystem storage for large tool results, removing in-memory size constraints on investigation outputs.
›Adds multi-cluster support and external change handling to Robusta tools.
›Adds GCP workload identity support for the GCP MCP integration.
›Adds OpenRouter as a supported AI provider.
└──▷ BREAKING ON UPGRADE
!Additional instructions support has been removed; any configuration relying on that feature will no longer work after upgrading.
!Unused Coralogix label settings have been removed and related functions have been flattened; configurations referencing those Coralogix label settings will break on upgrade.
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 v2.1.51 adds remote-control subcommand, custom npm registry support for plugins, and new SDK account environment variables.
└──▷ GET THIS VERSION
$ git clone --branch v2.1.51 https://github.com/anthropics/claude-code.git
# already have the repo? check out this version:$ git checkout v2.1.51
└──▷ TRY IT
Serve your local Claude Code environment to an external build system using the new remote-control subcommand.
$ claude remote-control
Extend the plugin git clone timeout when working behind a slow corporate proxy to avoid installation failures.
$ CLAUDE_CODE_PLUGIN_GIT_TIMEOUT_MS=300000 claude
Provide account metadata to Claude Code synchronously in SDK/CI contexts so telemetry events are correctly attributed from the very first event.
$ CLAUDE_CODE_ACCOUNT_UUID=<uuid> [email protected] CLAUDE_CODE_ORGANIZATION_UUID=<org-uuid> claude --print "<prompt>"
›Adds claude remote-control subcommand for external builds, enabling local environment serving for all users.
›Adds CLAUDE_CODE_PLUGIN_GIT_TIMEOUT_MS environment variable to configure plugin marketplace git timeout (default raised from 30s to 120s).
›Supports custom npm registries and specific version pinning when installing plugins from npm sources.
›Reduces the tool-result disk-persistence threshold from 100K to 50K characters, lowering context window consumption and extending conversation longevity.
›Adds CLAUDE_CODE_ACCOUNT_UUID, CLAUDE_CODE_USER_EMAIL, and CLAUDE_CODE_ORGANIZATION_UUID environment variables so SDK callers can supply account metadata synchronously, eliminating a race condition in early telemetry.
+2 moreshow less
›Routes HTTP hooks through the sandbox network proxy when sandboxing is enabled, enforcing the domain allowlist.
›The /model picker now shows human-readable labels (e.g., "Sonnet 4.5") instead of raw model IDs for pinned versions, with an upgrade hint when a newer version is available.
└──▷ BREAKING ON UPGRADE
!BashTool now skips the login shell (-l flag) by default when a shell snapshot is available (previously required setting CLAUDE_BASH_NO_LOGIN=true to achieve this); environments that relied on login-shell initialization running unconditionally may behave differently.
!HTTP hook header values no longer interpolate arbitrary environment variables — env var interpolation now requires an explicit allowedEnvVars list in the hook configuration, so existing hooks using env var interpolation in headers will stop expanding those variables.
Get up and running with Kimi-K2.6, GLM-5.2, MiniMax, DeepSeek, gpt-oss, Qwen, Gemma and other models.
Ollama v0.17.1 adds Nemotron architecture, web search for tool-capable models, and auto-update download controls.
└──▷ GET THIS VERSION
$ git clone --branch v0.17.1 https://github.com/ollama/ollama.git
# already have the repo? check out this version:$ git checkout v0.17.1
›Adds support for the Nemotron model architecture in Ollama's engine.
›Enables web search capabilities for models that support tools via the Ollama app.
›Adds configuration option to disable automatic update downloading.
›Improves LFM2 and LFM2.5 model support in Ollama's engine.
›MLX engine now delivers improved memory usage.
└──▷ BREAKING ON UPGRADE
!ollama create no longer defaults to affine quantization for unquantized models when using the MLX engine — existing workflows relying on that default will produce differently quantized output.
›Adds --nsa-prefill-backend and --nsa-decode-backend flags integrating TRT-LLM Native Sparse Attention kernels for DeepSeek V3.2, delivering 3x–5x throughput gains on Blackwell platforms.
›Adds native Anthropic API compatibility endpoint, enabling direct use of tools and clients built for the Anthropic API format.
›Adds SGLANG_MOONCAKE_CUSTOM_MEM_POOL environment variable for configuring Mooncake intra-node NVLink KV transfer memory pool.
›Adds routed-experts field to /completions and /chat/completions endpoints for MoE models.
›Introduces LoRA weight loading overlap with computation, reducing TTFT by ~78% and TPOT by ~34.88% on large adaptors.
+16 moreshow less
›Adds Flashinfer all-to-all MoE dispatcher for efficient expert parallelism communication in MoE models.
›Introduces FA4 (FP4 attention) backend and variable-length attention function for multimodal encoders, enabling lower-precision inference for vision-language models.
›Adds KV transfer support via MORI-IO for prefill-decode disaggregation.
›Adds INT4 kernels for the CPU backend.
›Adds JointThreshold algorithm for joint M2T and T2T decoding in diffusion LLMs.
›Adds basic dLLM scheduling strategy and implementation.
›Removes CUDA graph batch size limitation for diffusion LLMs.
›Updates AITER to v0.1.10.post3, adding FP8 Prefill, FP8 Decode, and FP8 KV Cache support on AMD hardware.
›Adds commit-to-version lookup tool at docs.sglang.io/references/release_lookup.html for tracing PRs to official releases.
›Adds new model support for Kimi-K2.5, GLM-5, Qwen 3.5, MiniMax 2.5, Ernie4.5-VL, Step3-VL, Step-3.5-Flash, LLaDA 2.1, Ring 2.5 1T / Ling 2.5 1T, MOVA (Diffusion), GLM-OCR, and DeepSeek-OCR-2.
›Adds diffusion pipeline support for parallel VAE decoding, token-level sequence sharding, Nunchaku and FP8 support, fused kernels, LTX-2, and multiple new models in the ComfyUI plugin.
›Adds MxINT4 Flashinfer TRT-LLM MoE GEMM support.
›Adds Kimi K2.5 Day 0 ROCm support and standardizes on ROCm 7.
›Adds Qwen 3.5 support on Ascend NPU.
›Adds MiniCPM3-4B model support on NPU/Ascend.
›Migrates GPTQ-Marlin and AWQ Marlin repack kernels to JIT compilation.
└──▷ BREAKING ON UPGRADE
!ROCm 6.3 is deprecated; ROCm 7 is now the standard for AMD hardware.
llama.cpp server adds max_completion_tokens request property as the replacement for deprecated max_tokens.
└──▷ GET THIS VERSION
$ git clone --branch b8144 https://github.com/ggml-org/llama.cpp.git
# already have the repo? check out this version:$ git checkout b8144
›Adds max_completion_tokens request property to the server API, setting the upper bound for reasoning+output tokens and superseding the now-deprecated max_tokens field.
Weaviate v1.36.0 promotes server-side batching, object TTL, backup restore cancellation, and inverted-index dropping to GA, and brings HFresh vector index into preview.
└──▷ GET THIS VERSION
$ git clone --branch v1.36.0 https://github.com/weaviate/weaviate.git
# already have the repo? check out this version:$ git checkout v1.36.0
└──▷ TRY IT
Back up only collections matching a wildcard pattern so nightly jobs don't require enumerating every class name.
$ POST /v1/backups/s3
{
"id": "nightly-2025-07-15",
"include": ["Product*", "Order*"]
}
›Adds wildcard support for include/exclude class lists in backup configuration, letting operators target collections by pattern.
›Introduces a debug abort endpoint for the object TTL subsystem (via PR #10543) to force-stop in-flight TTL deletion cycles.
›Exports new backup statuses in the OpenAPI/Swagger spec, enabling typed client integration with in-flight restore state.
›Object TTL reaches GA with batch deletions, throttled pause-every-X-batches cadence, Prometheus metrics, inactive-tenant handling, RBAC data delete permission enforcement, and schedule-based enforcement (TTL on a collection is only allowed when a schedule is configured).
›Alter Schema gains the ability to drop inverted indices from existing properties, with RBAC integration and multi-tenancy support, reclaiming disk space without recreating collections.
+3 moreshow less
›Adds VoyageAI V4 model support to the VoyageAI integration module.
›HNSW snapshots are now enabled by default, improving restart performance without manual configuration.
›Adds non-blocking segment deletions to reduce latency spikes during LSM compaction.
THE AI TOOLCHAIN