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 v0.6.57 adds Stripe billing, xAI Grok and Kimi K2.6 models, multi-platform server linking, and a flow-builder AI chat panel.
└──▷ GET THIS VERSION
$ git clone --branch autogpt-platform-beta-v0.6.57 https://github.com/Significant-Gravitas/AutoGPT.git
# already have the repo? check out this version:$ git checkout autogpt-platform-beta-v0.6.57
›Adds run_sub_session and AutoPilot block to queue follow-up messages on busy sessions via new UI support.
›Adds extract_context_messages utility and mode-gated --resume flag with compaction-aware gap-fill for unified transcript context.
›Adds TodoWrite tool for the baseline copilot.
›Adds platform server linking API for multi-platform CoPilot orchestration.
›Introduces subscription tier billing via Stripe Checkout, including deferred paid-to-paid downgrade and cancel-pending flow.
+13 moreshow less
›Adds xAI Grok 4.20 models available via OpenRouter.
›Adds Kimi K2.6 as fast default with a 4-config matrix, coalesced reasoning, and web_search tool support; SDK fast tier also defaults to Kimi K2.6 via OpenRouter with vendor-aware cost tracking.
›Adds Moonshot module with cache_control widening and partial-messages enabled by default.
›Enables live baseline streaming with a render flag, Sonar web_search integration, and simulator cost tracking.
›Streams extended_thinking on baseline via OpenRouter.
›Adds LaunchDarkly per-user model routing.
›Registers 13 new paid blocks and documents the credit/microdollar wallet boundary.
›Adds real OpenRouter cost tracking and cost-based rate limits (percent-only on public API).
›Adds Agent Briefing Panel surfacing scheduled agents in the library and copilot briefings, including computed monthly spend.
›Adds AI chat panel directly inside the flow builder.
›Adds system diagnostics and execution management dashboard.
›Adds zip download for agent outputs on the share page.
›Allows greater zoom-out range in the flow builder.
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 v2.1.117 adds fork-subagent support, richer OpenTelemetry attributes, and faster native search via embedded bfs/ugrep.
└──▷ GET THIS VERSION
$ git clone --branch v2.1.117 https://github.com/anthropics/claude-code.git
# already have the repo? check out this version:$ git checkout v2.1.117
└──▷ TRY IT
Surface custom/MCP slash-command names in OpenTelemetry traces for full audit visibility (they are redacted by default).
$ OTEL_LOG_TOOL_DETAILS=1 claude
›Enables forked subagents on external builds via CLAUDE_CODE_FORK_SUBAGENT=1.
›Loads agent frontmatter mcpServers for main-thread agent sessions launched with --agent.
›Persists /model selections across restarts even when a project pins a different model; startup header now indicates when the active model is from a project or managed-settings pin.
›Adds summarization offer in /resume for stale, large sessions before re-reading them.
›Concurrent MCP server connections now default when both local and claude.ai servers are configured, reducing startup time.
+9 moreshow less
›plugin install on an already-installed plugin now installs any missing dependencies instead of stopping.
›Adds claude plugin marketplace add auto-resolution of missing dependencies from configured marketplaces; dependency errors now include an install hint.
›Enforces managed-settings blockedMarketplaces and strictKnownMarketplaces on plugin install, update, refresh, and autoupdate.
›Advisor Tool (experimental) dialog now carries an 'experimental' label, a learn-more link, and a startup notification when enabled.
›Extends cleanupPeriodDays retention sweep to cover ~/.claude/tasks/, ~/.claude/shell-snapshots/, and ~/.claude/backups/.
›OpenTelemetry user_prompt events now include command_name and command_source for slash commands; cost.usage, token.usage, api_request, and api_error now include an effort attribute for models that support effort levels; custom/MCP command names are redacted unless OTEL_LOG_TOOL_DETAILS=1 is set.
›On native macOS and Linux builds, Glob and Grep tools are replaced by embedded bfs and ugrep available through the Bash tool, enabling faster searches without a separate tool round-trip.
›Windows builds now cache where.exe executable lookups per process for faster subprocess launches.
›Default effort for Pro/Max subscribers on Opus 4.6 and Sonnet 4.6 is now high (previously medium).
Autonomous coding agent as an SDK, IDE extension, or CLI assistant.
Cline v3.80.0 adds enterprise-managed skills, dynamic model recommendations, and a quota-exceeded error UI
└──▷ GET THIS VERSION
$ git clone --branch v3.80.0 https://github.com/cline/cline.git
# already have the repo? check out this version:$ git checkout v3.80.0
›Adds enterprise globalSkills support: remotely configured skills appear under a dedicated 'Enterprise Skills' section with toggle support, system prompt integration, and alwaysEnabled enforcement
›Onboarding flow now fetches recommended models dynamically instead of using a hardcoded list
›New dedicated 'Quota Exceeded' error message in chat UI when Cline account spend caps are reached
›Removes foreground terminal mode — all task command execution now runs in background mode, eliminating the VS Code integrated terminal dependency
└──▷ BREAKING ON UPGRADE
!Foreground terminal mode is removed; all task command execution now defaults to background mode, removing related settings UI — any workflow or configuration relying on foreground/VS Code integrated terminal execution will no longer work
Crush v0.62.0 cuts ~120k tokens/session with shorter tool descriptions and auto-renders diffs from any tool output.
└──▷ GET THIS VERSION
$ git clone --branch v0.62.0 https://github.com/charmbracelet/crush.git
# already have the repo? check out this version:$ git checkout v0.62.0
›Automatically detects and renders diff-formatted output from any tool (e.g. GitHub MCP server) with full diff UI, not just the built-in edit tool.
›Short tool descriptions are now the default, saving ~120k tokens per session and improving behaviour of small/local models with limited context windows.
└──▷ BREAKING ON UPGRADE
!Short tool descriptions are now the default; sessions relying on the verbose system prompt will behave differently unless CRUSH_SHORT_TOOL_DESCRIPTIONS=0 is explicitly set (this opt-out may be removed in a future release).
Return a final structured result from a tool without paying for an extra LLM turn.
typescript
return { result: myOutput, terminate: true };
›Adds ctx.ui.addAutocompleteProvider(...) so extensions can stack custom completion logic on top of built-in slash and path completion.
›Supports terminate: true on tool results, letting custom tools end a tool batch without triggering an automatic follow-up LLM call.
›Adds OSC 9;4 terminal progress indicators during agent streaming and compaction for terminals that support tab-bar activity display (iTerm2, WezTerm, Windows Terminal, Kitty).
›Migrates extension SDK to TypeBox 1.x with native tool argument validation that works in eval-restricted runtimes such as Cloudflare Workers.
└──▷ BREAKING ON UPGRADE
!Extensions and SDK integrations must now depend on and import from typebox 1.x instead of @sinclair/typebox 0.34.x; @sinclair/typebox/compiler is no longer shimmed.
!After ctx.newSession(), ctx.fork(), or ctx.switchSession(), pre-replacement session-bound extension objects (including captured pi and command ctx references) are invalidated and now throw instead of silently targeting the replaced session; post-switch work must be moved into the withSession callback using the provided ReplacedSessionContext.
›Adds mouse_wheel_zoom setting to enable changing editor font size via scroll wheel (hold ctrl on Linux/Windows, cmd on macOS); disabled by default.
›Adds limit_content_width setting in the agent panel to allow turning off the content max-width limit.
›Adds hover_popover_sticky and hover_popover_hiding_delay settings to control hover popover responsiveness.
›Adds cli_default_open_behavior setting to control whether zed <path> opens in an existing or new window; a first-run prompt saves the choice automatically.
›Adds editor: toggle block comment action, bound to cmd-k cmd-/ on macOS and ctrl-k ctrl-/ on Linux/Windows.
+8 moreshow less
›Adds workspace: format and save action that always formats regardless of format-on-save settings.
›Adds Claude Opus 4.7 as an available language model for BYOK (bring-your-own-key) configurations.
›Introduces parallel agents in Zed, enabling multiple concurrent agent threads.
›Adds anchor links for headings and footnote support in Markdown Preview.
›Adds syntax highlighting for TopoJSON files.
›Adds support for PNM image previews (.pbm, .ppm, .pgm files).
›Adds ability to toggle tree collapse in the Git panel.
›Reduces energy consumption for background windows by throttling their update speed.
Get up and running with Kimi-K2.6, GLM-5.2, MiniMax, DeepSeek, gpt-oss, Qwen, Gemma and other models.
Ollama v0.21.1 adds Kimi CLI integration, logprobs support in MLX, and faster MLX sampling.
└──▷ GET THIS VERSION
$ git clone --branch v0.21.1 https://github.com/ollama/ollama.git
# already have the repo? check out this version:$ git checkout v0.21.1
└──▷ TRY IT
Run Kimi K2.6 as an agentic assistant for long-horizon tasks directly from the CLI.
$ ollama launch kimi --model kimi-k2.6:cloud
›New ollama launch command installs and runs the Kimi CLI, enabling long-horizon agentic execution via a multi-agent system with models like kimi-k2.6:cloud.
›MLX runner now supports logprobs for compatible models.
›Faster MLX sampling via fused top-P and top-K in a single sort pass, with repeat penalties applied in the sampler.
›Improved MLX prompt tokenization by moving tokenization into request handler goroutines.
›GLM4 MoE Lite gains a performance improvement with a fused sigmoid router head.
WebGPU backend gains fused RMS_NORM + MUL kernel with optional GGML_WEBGPU_DISABLE_FUSION escape hatch.
└──▷ GET THIS VERSION
$ git clone --branch b8891 https://github.com/ggml-org/llama.cpp.git
# already have the repo? check out this version:$ git checkout b8891
›Adds fused rms_norm_mul kernel for the WebGPU backend, combining RMS normalisation and element-wise multiply into a single GPU dispatch to reduce overhead.
llama.cpp b8885 adds HunyuanVL vision-language model support with M-RoPE and GGUF conversion.
└──▷ GET THIS VERSION
$ git clone --branch b8885 https://github.com/ggml-org/llama.cpp.git
# already have the repo? check out this version:$ git checkout b8885
›Adds LLM_ARCH_HUNYUAN_VL architecture with M-RoPE (XD-RoPE) multi-dimensional position encoding support for HunyuanVL image tokens.
›Adds PROJECTOR_TYPE_HUNYUANVL with PatchMerger vision encoder for HunyuanVL multimodal inference.
›Adds GGUF conversion support for HunyuanVL vision and text models (HunyuanVL and HunyuanOCR) via convert_hf_to_gguf.py, verified with F16 and Q8_0 quantization on Metal.
Phoenix v14.11.0 adds session summary evals, a secrets settings page, and a root span name in session turn lists.
└──▷ GET THIS VERSION
$ git clone --branch arize-phoenix-v14.11.0 https://github.com/Arize-ai/phoenix.git
# already have the repo? check out this version:$ git checkout arize-phoenix-v14.11.0
›Adds a session summary eval and sidebar-title prompt rewrite capability for agent sessions.
›Adds a Secrets settings page for managing secrets directly in the Phoenix UI.
›Shows root span name in the session turn list for easier agent session navigation.
›Adds a 'Show Table Aside' toggle to the project filter configuration (feature-flagged).
Phoenix client passes trace_id to experiment evaluators and adds type-aware attribute filtering to GET /v1/spans.
└──▷ GET THIS VERSION
$ git clone --branch arize-phoenix-client-v2.4.0 https://github.com/Arize-ai/phoenix.git
# already have the repo? check out this version:$ git checkout arize-phoenix-client-v2.4.0
›Adds type-aware attribute filter to GET /v1/spans REST endpoint, enabling more precise span queries that match on attribute type as well as value.
›Passes trace_id through to experiment evaluators in the client, making trace correlation available inside evaluator logic.
THE AI TOOLCHAIN