Backfill issue -272 — The AI Toolchain

EDITIONS tail grep head diff uniq

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.

VIEW

ISSUE VIEW full issue

Do you prefer this view?

$ tct list   # 10 tools matched

AI & LLM Tooling

◆ AI Coding Agents

Cline

v3.38.0 NOTES

Autonomous coding agent as an SDK, IDE extension, or CLI assistant.

Cline v3.38.0 adds Gemini 3.0 Pro support with thought signatures, Bedrock context-window retry handling, and remote dynamic onboarding model lists.

└──▷ GET THIS VERSION

$ git clone --branch v3.38.0 https://github.com/cline/cline.git
# already have the repo? check out this version:
$ git checkout v3.38.0

›Adds Gemini 3.0 Pro to the Gemini provider and featured/onboarding model lists.
›Adds thought signature support for the Gemini SDK, enabling reasoning transparency.
›Adds context window error detection and automatic retry handling for AWS Bedrock.
›Supports feature flag payloads and a remote dynamic onboarding model list.
›Upgrades SAP AI SDK JS packages to a new major version.

└──▷ BREAKING ON UPGRADE

!Firebase authentication has been retired; setups relying on Firebase-based auth will break on upgrade.

Was this useful?

Continue

@continuedev/[email protected] NOTES

Continue 1.29.0 adds Granite 4 FIM and tool calling, a report failure tool, and CLI context percentage display.

└──▷ GET THIS VERSION

$ git clone --branch @continuedev/[email protected] https://github.com/continuedev/continue.git
# already have the repo? check out this version:
$ git checkout @continuedev/[email protected]

›Adds a report failure tool, filterable via --id flag, giving agents a structured way to signal task failure.
›CLI now always displays the context window usage percentage.
›CLI skips checking for updates when running in dev mode.
›Adds support for Granite 4 fill-in-the-middle (FIM) and tool calling.
›MCP warning messages now surface which MCPs are configured.

+3 moreshow less

›Stream error toggle now stays open by default for easier debugging.
›Applied rules are no longer included in chat history or dev data.
›Thinking block UI redesigned to occupy less screen space.

Was this useful?

SST OpenCode

v1.0.76 2 RELEASES · 2025-11-18 NOTES STABLE

The open source coding agent.

OpenCode v1.0.76 adds opencode agent list to browse all available agents with details.

└──▷ GET THIS VERSION

$ git clone --branch v1.0.76 https://github.com/sst/opencode.git
# already have the repo? check out this version:
$ git checkout v1.0.76

└──▷ TRY IT

Discover which agents are configured and available before starting a session.

$ opencode agent list

›New opencode agent list command displays all available agents with their details.

1 more release in this issue · 2025-11-18

v1.0.73 NOTES STABLE

OpenCode v1.0.73 adds Nix installation support and Swift syntax highlighting.

└──▷ GET THIS VERSION

$ git clone --branch v1.0.73 https://github.com/sst/opencode.git
# already have the repo? check out this version:
$ git checkout v1.0.73

›Adds Nix support for installing and managing OpenCode as a package.
›Adds Swift syntax highlighting support in the editor/output.
›Reworked default model selection.

└──▷ BREAKING ON UPGRADE

!scroll_speed now only accepts positive integers; non-integer positive numbers that were previously valid are no longer accepted.

Was this useful?

Zed

v0.212.7 NOTES

Zed v0.212.7 adds thought signatures for Gemini tool calls.

└──▷ GET THIS VERSION

$ git clone --branch v0.212.7 https://github.com/zed-industries/zed.git
# already have the repo? check out this version:
$ git checkout v0.212.7

›Adds thought signatures for Gemini tool calls, surfacing model reasoning inline during AI-assisted workflows.

Was this useful?

Google gemini-cli

v0.18.0-nightly.20251118.7cc5234b9 2 RELEASES · 2025-11-18 NOTES STABLE

An open-source AI agent that brings the power of Gemini directly into your terminal.

gemini-cli v0.18.0-nightly adds Gemini 3, bulk extension uninstall, reworked MCP tool discovery, and right-click paste in Alternate Buffer mode.

└──▷ GET THIS VERSION

$ git clone --branch v0.18.0-nightly.20251118.7cc5234b9 https://github.com/google-gemini/gemini-cli.git
# already have the repo? check out this version:
$ git checkout v0.18.0-nightly.20251118.7cc5234b9

›Launches Gemini 3 model support in Gemini CLI.
›Enables uninstalling multiple extensions in a single command.
›Reworks MCP tool discovery and invocation for improved reliability.
›Adds right-click to paste in Alternate Buffer mode.
›Shows a tip on the first request to guide new users.

1 more release in this issue · 2025-11-18

v0.16.0 NOTES STABLE

Gemini CLI v0.16.0 launches Gemini 3 model support.

└──▷ GET THIS VERSION

$ git clone --branch v0.16.0 https://github.com/google-gemini/gemini-cli.git
# already have the repo? check out this version:
$ git checkout v0.16.0

›Adds Gemini 3 model support directly within Gemini CLI.

Was this useful?

◆ AI Agent Frameworks

OpenAI Agents SDK

v0.6.0 NOTES

OpenAI Agents SDK v0.6.0 adds parallel input guardrails, prompt cache retention, tool error logging, and a breaking handoff history change.

└──▷ GET THIS VERSION

$ git clone --branch v0.6.0 https://github.com/openai/openai-agents-python.git
# already have the repo? check out this version:
$ git checkout v0.6.0

└──▷ USE IT

Pin a prompt cache retention window in ModelSettings to control how long cached prompts are retained for a cost- or latency-sensitive agent.

python

from agents import Agent, ModelSettings

agent = Agent(
    name='my-agent',
    model='gpt-4o',
    model_settings=ModelSettings(prompt_cache_retention=300),
)

›Adds prompt_cache_retention field to ModelSettings to control prompt cache retention behaviour.
›Adds run_in_parallel parameter to input guardrails, allowing multiple guardrails to execute concurrently instead of sequentially.
›Adds tool error logging so errors raised during tool execution are now captured in logs.
›Handoff message history is now collapsed into a single message by default when handing off to a new agent (replaces the previous multi-message history pass-through).

└──▷ BREAKING ON UPGRADE

!On agent handoff, message history is now collapsed into a single message by default ('Nest handoff history by default'). Agents that previously relied on the full expanded message history being passed to the receiving agent may behave differently; test before upgrading to v0.6.0 in production.

Was this useful?

PydanticAI

v1.19.0 NOTES

PydanticAI v1.19.0 adds metadata passthrough to deferred tool exceptions and Anthropic token-counting support.

└──▷ GET THIS VERSION

$ git clone --branch v1.19.0 https://github.com/pydantic/pydantic-ai.git
# already have the repo? check out this version:
$ git checkout v1.19.0

└──▷ USE IT

Attach routing or audit metadata to a deferred tool call so downstream handlers know what to do with it.

python

from pydantic_ai.exceptions import CallDeferred

raise CallDeferred(metadata={'queue': 'human-review', 'priority': 'high'})

›Adds count_tokens method to AnthropicModel for explicit token counting.
›Adds support for UsageLimits.count_tokens_before_request with AnthropicModel, enabling pre-flight token budget checks.
›Allows metadata to be passed to CallDeferred and ApprovalRequired exceptions, propagating it onto DeferredToolRequests.

Was this useful?

browser-use

0.9.7 NOTES

browser-use 0.9.7 adds Gemini 3 model support and configurable screenshot sizing.

└──▷ GET THIS VERSION

$ git clone --branch 0.9.7 https://github.com/browser-use/browser-use.git
# already have the repo? check out this version:
$ git checkout 0.9.7

›Adds support for Gemini 3 as a usable model.
›Enables configurable screenshot size for browser-use models.

Was this useful?

◆ Local LLM Runtimes

vLLM

v0.11.1 NOTES

vLLM v0.11.1 adds Anthropic /v1/messages API, DeepSeek-V3.2 model, GCS streamer support, and async scheduling improvements across 1456 commits.

└──▷ GET THIS VERSION

$ git clone --branch v0.11.1 https://github.com/vllm-project/vllm.git
# already have the repo? check out this version:
$ git checkout v0.11.1

└──▷ TRY IT

Query vLLM using an Anthropic-compatible client against the new /v1/messages endpoint.

$ curl http://localhost:8000/v1/messages \
  -H 'Content-Type: application/json' \
  -H 'x-api-key: <your-api-key>' \
  -d '{"model": "deepseek-ai/DeepSeek-V3", "max_tokens": 256, "messages": [{"role": "user", "content": "Hello!"}]}'

Dump torch.compile debug artifacts to disk for inspecting graph partitions and compile-cache issues.

$ VLLM_DEBUG_DUMP_PATH=/tmp/vllm_compile_debug vllm serve meta-llama/Llama-3-8B --compilation-config '{"cudagraph_mode": "full_graph"}'

›Adds VLLM_DEBUG_DUMP_PATH environment variable for dumping torch.compile debug artifacts.
›Updates default CUDA build to torch==2.9.0+cu129 (PyTorch 2.9.0 + CUDA 12.9.1), enabling Inductor partitioning.
›Adds --async-scheduling flag with robustness fixes covering chunked prefill, structured outputs, priority scheduling, MTP, and DeepEP/DCP; expected to become the default in the next release.
›Adds support for the DeepSeek-V3.2 model.
›Adds Run:ai model streamer GCS package support.

+18 moreshow less

›Adds media domain restriction filtering via add option to restrict media domains, preventing bypass via HTTP redirects.
›Adds RL online quantization support with torchao.
›Adds cuda2cpu support in NixlConnector for P/D disaggregated serving.
›Adds KV load failure recovery support in V1 P/D disaggregated mode.
›Adds MLA cache support with different latent dimensions in NIXL.
›Adds explicit pooling classes for the Transformers backend.
›Adds Eagle/Eagle3 multimodal support for speculative decoding, with enablement on Qwen2.5-VL.
›Adds text-only mode for Qwen3-VL multimodal model.
›Adds LoRA support for OPT models.
›Adds ROCm support for AMD Ryzen AI MAX / AI 300 Series GPUs.
›Adds Blackwell GPU family support (SM100+) including INT8 quantization fixes.
›Adds batch-invariant torch.compile support across attention and MoE backends, with explicit support for DeepGEMM and FlashInfer on Hopper and Blackwell GPUs.
›Updates to Transformers v4.56.2.
›Adds chunk-aligned Mamba2 kernel.
›Adds GC debug callback for core diagnostics.
›Adds more tuned Triton configs for ExpertsInt8 and FP8 MoE kernels.
›Serializes cudagraph_mode as its enum name instead of numeric value in torch.compile.
›Adds chat template kwargs filtering for API security.

└──▷ BREAKING ON UPGRADE

!The --endpoint-type benchmark flag, deprecated in v0.11.0, is now removed.
!VllmConfig is moved from config/__init__.py to config/vllm.py; imports that referenced the old path will break.
!vllm.worker module is removed; any code importing from vllm.worker directly will break.

Was this useful?

Other / Uncategorized

◆ VECTOR DB RAG

Chroma

1.3.5 NOTES

Chroma 1.3.5 adds Nomic and Google GenAI embedding functions, Transformers.js EF, GCS storage, keepalive/max-conn controls, and non-prefixed EF env vars.

└──▷ GET THIS VERSION

$ git clone --branch 1.3.5 https://github.com/chroma-core/chroma.git
# already have the repo? check out this version:
$ git checkout 1.3.5

›Adds support for non-prefixed embedding-function environment variables in the Python client, so EF config env vars no longer require a prefix.
›Adds keepalive and maximum connections configuration to the Python client for tuning long-lived gRPC/HTTP connections.
›Adds a Nomic embedding function to the Python client via the [ENH] Add nomic embedding function integration.
›Adds a Google GenAI embedding function to the Python client.
›Adds a Transformers.js embedding function to the JavaScript client, with compatibility maintained with the Python client.

+4 moreshow less

›Adds auto-loading of the embedding-function package if it is already installed, removing the need to manually import it.
›Adds a GCS (Google Cloud Storage) client as a storage backend.
›Adds schema validation for embedding functions defined in a collection schema, with client-side validation that a sparse source key requires an explicit embedding function.
›Garbage-collects soft-deleted attached functions to reclaim storage over time.

Was this useful?

// END OF TRANSMISSION

The AI Toolchain · curated daily · November 18, 2025
Releases & features from the open-source and commercial AI tools that practitioners actually run.

The AI Toolchain — issue -272, November 18, 2025