Backfill issue -162 — 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

GitHub Copilot CLI

v1.0.3 NOTES

Copilot CLI v1.0.3 adds MCP config from devcontainer.json, --binary-version flag, /restart command, Windows Terminal support, and more.

└──▷ GET THIS VERSION

$ git clone --branch v1.0.3 https://github.com/github/copilot-cli.git
# already have the repo? check out this version:
$ git checkout v1.0.3

└──▷ USE IT

Update the renamed merge strategy key in your project config to avoid unknown-field warnings after upgrading.

yaml

mergeStrategy: squash

›Adds --binary-version flag to query the CLI binary version without launching a full session.
›Reads MCP server configuration from .devcontainer/devcontainer.json.
›Adds extraKnownMarketplaces repository setting to replace marketplace definitions.
›Adds /restart command to hot restart the CLI while preserving the current session.
›Extends /terminal-setup command with Windows Terminal support.

+12 moreshow less

›Documents GH_HOST, HTTP_PROXY, HTTPS_PROXY, NO_COLOR, and NO_PROXY environment variables in help output.
›The mergeStrategy config key replaces the former merge_strategy key (renamed for consistency).
›Enables alternate screen buffer by default for staff users.
›Makes Extensions available as an experimental feature, allowing custom tools and hooks via @github/copilot-sdk.
›Type quit to exit the CLI, in addition to the existing exit command.
›/reset-allowed-tools now fully undoes /allow-all and re-triggers the autopilot permission dialog.
›MCP servers with non-conforming outputSchema are now accessible.
›/add-dir directories persist across session changes such as /clear and /resume.
›Background task notifications display in the timeline with expandable detail.
›Retry status messages display to show progress during server error recovery.
›Loading spinner shown in diff mode while fetching changes.
›/init suggestion is suppressed when .github/instructions/ already contains instructions.

└──▷ BREAKING ON UPGRADE

!The merge_strategy config key is renamed to mergeStrategy; existing configs using merge_strategy must be updated.

Was this useful?

SST OpenCode

v1.2.24 NOTES

The open source coding agent.

OpenCode v1.2.24 adds workspace support in TUI, GitLab 1M context window, and Copilot GPT-5.4 xhigh.

└──▷ GET THIS VERSION

$ git clone --branch v1.2.24 https://github.com/sst/opencode.git
# already have the repo? check out this version:
$ git checkout v1.2.24

›Adds initial workspace support in the TUI for multi-project navigation.
›Enables GitLab's 1M context window by sending the context-1m-2025-08-07 beta header.
›Adds Copilot GPT-5.4 xhigh model support.

Was this useful?

Alibaba Qwen Code

v0.12.0 NOTES

Qwen Code v0.12.0 adds a hooks system, GIF terminal capture, PTY-by-default shell, AskUserQuestionTool, and richer MCP/extension TUIs.

└──▷ GET THIS VERSION

$ git clone --branch v0.12.0 https://github.com/QwenLM/qwen-code.git
# already have the repo? check out this version:
$ git checkout v0.12.0

›Implements a hooks system infrastructure (hookConfig) with CLI and UI management, including enable and max turns configuration for lifecycle events such as userPromptSubmit.
›Adds tabWidth support for code highlighting in CodeColorizer, replacing tabs with spaces at a configurable width.
›Enables PTY by default for shell execution, with various shell enhancements.
›Adds streaming terminal capture with GIF generation.
›Adds custom QC commands for GitHub workflows.

+6 moreshow less

›Adds an export-HTML viewer for tool call results.
›Adds JSON Schema validation for VS Code settings.
›Implements AskUserQuestionTool for interactive user queries during agent runs.
›Enhances the MCP Management TUI with dynamic enable/disable and runtime updates.
›Adds an interactive TUI for extension management.
›Adds support for printable CSI-u keys in KeypressContext.

Was this useful?

◆ AI Agent Frameworks

LangChain

langchain-openai==1.1.11 NOTES

langchain-openai 1.1.11 adds tool search support and streaming token usage for OpenRouter.

└──▷ GET THIS VERSION

$ git clone --branch langchain-openai==1.1.11 https://github.com/langchain-ai/langchain.git
# already have the repo? check out this version:
$ git checkout langchain-openai==1.1.11

›Adds tool search support via the OpenAI integration (feat(openai): support tool search).
›Adds streaming token usage support for OpenRouter.

Was this useful?

LangChain LangGraph

sdk==0.3.10 NOTES

Build resilient agents.

LangGraph SDK 0.3.10 adds type-safe stream/invoke with proper output type coercion

└──▷ GET THIS VERSION

$ git clone --branch sdk==0.3.10 https://github.com/langchain-ai/langgraph.git
# already have the repo? check out this version:
$ git checkout sdk==0.3.10

›Adds type-safe stream/invoke calls with proper output type coercion, enabling strongly-typed responses from graph runs

Was this useful?

OpenAI Agents SDK

v0.11.0 NOTES

OpenAI Agents SDK v0.11.0 adds tool search support with namespaces and extends computer use to the GA gpt-5.4 model.

└──▷ GET THIS VERSION

$ git clone --branch v0.11.0 https://github.com/openai/openai-agents-python.git
# already have the repo? check out this version:
$ git checkout v0.11.0

›Adds tool search tool support via the Responses API, including namespace scoping — see the tool_search.py example for usage.
›Extends ComputerTool to support the GA gpt-5.4 model in addition to the existing computer-use-preview model.

Was this useful?

OpenClaw

v2026.3.8 NOTES

Your own personal AI assistant. Any OS.

OpenClaw v2026.3.8 adds local backup commands, ACP provenance tracking, Brave LLM Context search, and configurable Talk silence timeout.

└──▷ GET THIS VERSION

$ git clone --branch v2026.3.8 https://github.com/openclaw/openclaw.git
# already have the repo? check out this version:
$ git checkout v2026.3.8

└──▷ TRY IT

Create a config-only backup before a destructive operation such as a major upgrade or workspace reset.

$ openclaw backup create --only-config

Verify the integrity of an existing local state archive to confirm the manifest and payload are intact.

$ openclaw backup verify

Enable Brave LLM Context mode so web_search returns grounding snippets with source metadata instead of raw results.

yaml

tools.web.search.brave.mode: "llm-context"

›Adds openclaw backup create and openclaw backup verify commands for local state archives, with --only-config and --no-include-workspace flags, manifest/payload validation, and backup guidance in destructive flows.
›Adds openclaw acp --provenance off|meta|meta+receipt flag so agents can retain and report ACP-origin context with session trace IDs.
›Adds opt-in tools.web.search.brave.mode: "llm-context" config to call Brave's LLM Context endpoint and return extracted grounding snippets with source metadata.
›Adds top-level talk.silenceTimeoutMs config so Talk mode waits a configurable amount of silence before auto-sending the current transcript.
›Includes the short git commit hash in openclaw --version output when metadata is available.

+3 moreshow less

›Adds a remote gateway token field for macOS onboarding remote mode, preserving existing non-plaintext gateway.remote.token config values until explicitly replaced.
›TUI now infers the active agent from the current workspace when launched inside a configured agent workspace, while preserving explicit agent: session targets.
›Adds browser.relayBindHost config so the Chrome relay can bind to an explicit non-loopback address for WSL2 and other cross-namespace setups.

Was this useful?

◆ Local LLM Runtimes

llama.cpp

b8248 2 RELEASES · 2026-03-09 NOTES STABLE

llama.cpp b8248 displays total and free VRAM capacity during CUDA device initialization.

└──▷ GET THIS VERSION

$ git clone --branch b8248 https://github.com/ggml-org/llama.cpp.git
# already have the repo? check out this version:
$ git checkout b8248

›Displays total and free VRAM capacity during CUDA device initialization, giving practitioners immediate visibility into GPU memory availability at startup.

1 more release in this issue · 2026-03-09

b8247 NOTES STABLE

llama-bench gains -hf and -hff flags for Hugging Face model sourcing, with --mmap 1 now default.

└──▷ GET THIS VERSION

$ git clone --branch b8247 https://github.com/ggml-org/llama.cpp.git
# already have the repo? check out this version:
$ git checkout b8247

└──▷ TRY IT

Benchmark a model pulled directly from Hugging Face without manually downloading it first.

$ llama-bench -hf <org>/<model-repo>

›Adds -hf and -hff flags to llama-bench for specifying models directly from Hugging Face.
›llama-bench now defaults to --mmap 1 (memory-mapped file I/O enabled by default).

└──▷ BREAKING ON UPGRADE

!llama-bench now uses --mmap 1 by default; existing benchmarks that relied on the previous default (mmap off) may see different memory or performance characteristics without explicitly setting --mmap 0.

Was this useful?

Other / Uncategorized

◆ AI OBSERVABILITY

Arize Phoenix

arize-phoenix-v13.12.0 NOTES

Adds parent_id filter to GET spans endpoints for targeted span tree queries.

└──▷ GET THIS VERSION

$ git clone --branch arize-phoenix-v13.12.0 https://github.com/Arize-ai/phoenix.git
# already have the repo? check out this version:
$ git checkout arize-phoenix-v13.12.0

└──▷ TRY IT

Retrieve only the child spans of a specific parent span to investigate a single trace subtree.

$ GET /v1/spans?parent_id=<span_id>

›Adds parent_id filter to GET spans endpoints, enabling queries scoped to child spans of a specific parent span.
›Adds a feature-flagged tracing onboarding empty state to guide new users through tracing setup.
›Adds PackageManagerCommandBlock UI component for displaying package manager commands.

Was this useful?

◆ VECTOR DB RAG

LanceDB

v0.27.0-beta.5 4 RELEASES · 2026-03-09 NOTES STABLE

LanceDB v0.27.0-beta.5 adds JS native array inference and upgrades Lance to 3.0.0-rc.3.

└──▷ GET THIS VERSION

$ git clone --branch v0.27.0-beta.5 https://github.com/lancedb/lancedb.git
# already have the repo? check out this version:
$ git checkout v0.27.0-beta.5

›Adds automatic inference of JavaScript native arrays, removing the need to manually specify array schema when ingesting JS data.
›Upgrades the underlying Lance storage engine to 3.0.0-rc.3.

3 more releases in this issue · 2026-03-09

python-v0.30.0-beta.5 NOTES STABLE

LanceDB python-v0.30.0-beta.5 adds JS native array inference and upgrades Lance to 3.0.0-rc.3.

└──▷ GET THIS VERSION

$ git clone --branch python-v0.30.0-beta.5 https://github.com/lancedb/lancedb.git
# already have the repo? check out this version:
$ git checkout python-v0.30.0-beta.5

›Adds automatic inference of JavaScript native arrays, reducing manual schema specification when ingesting JS-native data.
›Upgrades the Lance backend to version 3.0.0-rc.3.

v0.27.0-beta.4 NOTES STABLE

LanceDB v0.27.0-beta.4 adds num_deleted_rows to delete results, remote index params, and dict-to-SQL struct conversion in table.update()

└──▷ GET THIS VERSION

$ git clone --branch v0.27.0-beta.4 https://github.com/lancedb/lancedb.git
# already have the repo? check out this version:
$ git checkout v0.27.0-beta.4

└──▷ USE IT

Confirm how many rows were removed after a delete operation using the new num_deleted_rows field.

python

result = table.delete("category = 'obsolete'")
print(result.num_deleted_rows)

Update a row with a nested struct column by passing a plain Python dict instead of manually constructing SQL.

python

table.update(where="id = 42", values={"metadata": {"source": "ingest", "version": 3}})

›Adds num_deleted_rows field to the delete operation result, letting callers confirm how many rows were removed.
›Adds support for remote index params, enabling index configuration through the remote SDK.
›Adds parity for the fast_search keyword argument between vector and FTS searches.
›Supports dict-to-SQL struct conversion in table.update() (Python), simplifying structured updates without manual SQL construction.
›Allows passing Azure client ID and tenant ID through the remote SDK for Azure-backed connections.

python-v0.30.0-beta.4 NOTES STABLE

LanceDB python-v0.30.0-beta.4 adds delete result row counts, remote index params, fast_search parity for FTS, dict-to-struct in update(), and Azure client/tenant ID passthrough.

└──▷ GET THIS VERSION

$ git clone --branch python-v0.30.0-beta.4 https://github.com/lancedb/lancedb.git
# already have the repo? check out this version:
$ git checkout python-v0.30.0-beta.4

└──▷ USE IT

Update a struct-typed column by passing a plain Python dict instead of manually constructing an Arrow struct.

python

table.update(where="id = 42", values={"metadata": {"source": "upload", "version": 2}})

›Adds num_deleted_rows field to the result returned by table.delete(), letting callers inspect how many rows were removed.
›Adds support for fast_search keyword argument in full-text search (FTS) queries, bringing parity with vector search.
›Supports dict-to-SQL-struct conversion in table.update(), allowing Python dicts to be passed directly as struct values.
›Allows passing Azure client ID and tenant ID through the remote SDK when connecting to Azure-backed LanceDB deployments.
›Adds support for remote index params, enabling index configuration to be specified via the remote SDK.

Was this useful?

// END OF TRANSMISSION

The AI Toolchain · curated daily · March 9, 2026
Releases & features from the open-source and commercial AI tools that practitioners actually run.

The AI Toolchain — issue -162, March 9, 2026