Backfill issue -217 — 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 Agent Frameworks

Agno (formerly Phidata)

v2.3.25 NOTES

Agno v2.3.25 adds LearningMachine for per-interaction agent learning and an AST-based CodeChunking strategy.

└──▷ GET THIS VERSION

$ git clone --branch v2.3.25 https://github.com/agno-agi/agno.git
# already have the repo? check out this version:
$ git checkout v2.3.25

›Adds LearningMachine, a unified learning system that coordinates multiple learning types — each with its own storage backend and retrieval pattern — so agents can learn from every interaction.
›Adds CodeChunking strategy that uses ASTs to split code into contextually relevant segments, complementing existing text-based chunkers.

Was this useful?

LangChain LangGraph

1.0.6 2 RELEASES · 2026-01-12 NOTES STABLE

Build resilient agents.

LangGraph 1.0.6 adds compile-time checkpointer type validation to catch configuration errors early.

└──▷ GET THIS VERSION

$ git clone --branch 1.0.6 https://github.com/langchain-ai/langgraph.git
# already have the repo? check out this version:
$ git checkout 1.0.6

›Validates checkpointer type at compile time, surfacing misconfigured checkpointers before runtime.

1 more release in this issue · 2026-01-12

prebuilt==1.0.6 NOTES STABLE

LangGraph prebuilt 1.0.6 adds compile-time checkpointer validation, custom encryption at rest, and paginated assistant search.

└──▷ GET THIS VERSION

$ git clone --branch prebuilt==1.0.6 https://github.com/langchain-ai/langgraph.git
# already have the repo? check out this version:
$ git checkout prebuilt==1.0.6

›Validates checkpointer type at compile time, catching misconfiguration before runtime.
›Supports custom encryption at rest for checkpoint data.
›Includes pagination in assistants search responses.

Was this useful?

OpenClaw

v2026.1.11 NOTES

Your own personal AI assistant. Any OS.

OpenClaw v2026.1.11 adds a first-class plugin system, voice-call plugin, apply_patch tool, $include config, and cron --model flag.

└──▷ GET THIS VERSION

$ git clone --branch v2026.1.11 https://github.com/openclaw/openclaw.git
# already have the repo? check out this version:
$ git checkout v2026.1.11

└──▷ TRY IT

Install a third-party plugin from a local tarball and verify it loads correctly.

$ clawdbot plugins install ./my-plugin.tgz && clawdbot plugins doctor

Split a large config into separate files to keep provider and agent settings maintainable.

yaml

# clawdbot.config.yaml
$include: ./providers.yaml
$include: ./agents.yaml

Schedule a cron task with a specific model override so high-stakes jobs use a more capable model.

$ clawdbot cron add --model minimax/m2.1 "0 9 * * 1" "Summarise last week's alerts and post to #security"

›New plugin extension loader supporting tools, RPC, CLI, and services — with discovery paths, config schema, and Control UI labels (uiHints).
›New clawdbot plugins install command (path/tgz/npm) plus list, info, enable, disable, and doctor subcommands for full plugin lifecycle management.
›New Voice Call plugin (Twilio/log backend) with start/status RPC, CLI, and tool interfaces.
›New apply_patch tool for multi-file edits (experimental; gated by tools.exec.applyPatch; OpenAI-only).
›Adds $include directive for modular, split config files.

+13 moreshow less

›Adds --model flag to cron add and cron edit commands.
›New pre-compaction memory flush config (agents.defaults.compaction.*) with soft threshold and system prompt to persist durable memories before auto-compaction.
›Per-DM history limit overrides via dmHistoryLimit at the provider level.
›Discord channel and category management actions now exposed in the message tool.
›Gateway now requires client.id in WebSocket connect params; uses client.instanceId for presence de-duplication.
›Control UI Config tab adds model presets (MiniMax M2.1, GLM 4.7, Kimi) for one-click provider setup.
›Auto-reply compact /model picker now shows models alongside available providers and displays provider endpoints in /model status.
›Bundles skill-creator skill to guide creating and packaging skills.
›Installer supports CLAWDBOT_NO_ONBOARD=1 for non-interactive installs and --install-method git|npm flag with auto-detection of source checkouts.
›Installer adds install.sh --help with flags, env vars, and git install hint.
›macOS: prompts to install the global clawdbot CLI when missing in local mode, installing via clawd.bot/install-cli.sh.
›Onboarding adds Moonshot AI (Kimi K2) auth choice and MiniMax M2.1 Lightning auth choice.
›CLI/Onboarding prompts to reuse detected API keys for Moonshot, MiniMax, Z.AI, Gemini, Anthropic, and OpenCode.

└──▷ BREAKING ON UPGRADE

!The bash tool is renamed to exec (a config alias is maintained, but tooling referencing the old name bash should be verified).
!Gateway now requires client.id in WebSocket connect params; existing clients omitting this field will fail to connect.
!macOS local mode always manages launchd; the attach-only gateway setting has been removed.

Was this useful?

◆ AI Coding Agents

Charm Crush

v0.32.0 NOTES

Glamourous agentic coding for all

Crush v0.32.0 adds a config option to disable default providers and allows sending prompts with attachments but no text.

└──▷ GET THIS VERSION

$ git clone --branch v0.32.0 https://github.com/charmbracelet/crush.git
# already have the repo? check out this version:
$ git checkout v0.32.0

└──▷ TRY IT

Suppress all default cloud providers so only locally configured models are available — useful in air-gapped or local-only setups.

$ CRUSH_DISABLE_DEFAULT_PROVIDERS=1 crush

›Adds disable_default_providers config setting to suppress all built-in cloud providers, useful when running exclusively with local models.
›Enables sending a prompt that contains only file/text attachments with no typed text.

Was this useful?

SST OpenCode

v1.1.15 2 RELEASES · 2026-01-12 NOTES STABLE

The open source coding agent.

OpenCode v1.1.15 adds password authentication, wildcard permission matching, and TUI hints for variant toggling.

└──▷ GET THIS VERSION

$ git clone --branch v1.1.15 https://github.com/sst/opencode.git
# already have the repo? check out this version:
$ git checkout v1.1.15

└──▷ TRY IT

Secure your OpenCode server with a password so only authenticated clients can connect.

$ OPENCODE_SERVER_PASSWORD=mysecret opencode

›Adds password authentication to improve server security.
›Supports wildcard permissions matching commands both with and without arguments.
›Adds a hint for the variants toggle in the TUI.
›Redesigns tips display on the home screen.

└──▷ BREAKING ON UPGRADE

!The OPENCODE_PASSWORD environment variable is renamed to OPENCODE_SERVER_PASSWORD; existing setups using OPENCODE_PASSWORD will no longer authenticate correctly until updated.

1 more release in this issue · 2026-01-12

v1.1.14 NOTES STABLE

OpenCode v1.1.14 adds MCP resource content type support, mouse interactivity in TUI, and image previews in desktop.

└──▷ GET THIS VERSION

$ git clone --branch v1.1.14 https://github.com/sst/opencode.git
# already have the repo? check out this version:
$ git checkout v1.1.14

›Supports resource content type in MCP tool output, enabling richer structured data from MCP tools.
›Adds mouse hover and click support to the questions tool in the TUI.
›Adds mouse hover and click support to the TUI autocomplete.
›Adds clickable navigation to the subagent header in the TUI.
›Copies OAuth URL to clipboard when device code is unavailable during authentication.

+1 moreshow less

›Adds image preview support for image attachments in the desktop app.

Was this useful?

Earendil Works Pi

v0.44.0 2 RELEASES · 2026-01-12 NOTES STABLE

AI agent toolkit: unified LLM API, agent loop, TUI, coding agent CLI

Pi v0.44.0 adds session naming, fuzzy settings search, and page-jump navigation in the session selector.

└──▷ GET THIS VERSION

$ git clone --branch v0.44.0 https://github.com/earendil-works/pi.git
# already have the repo? check out this version:
$ git checkout v0.44.0

└──▷ TRY IT

Give the current session a meaningful name so forked sessions are easy to tell apart in the session selector.

$ /name threat-hunting-2024-q2

›New /name <name> command sets a display name for the current session in the session selector, useful for distinguishing forked sessions.
›Extensions can programmatically get and set session names via new pi.setSessionName() and pi.getSessionName() APIs.
›New notify.ts extension example demonstrates desktop notifications via the OSC 777 escape sequence.
›Page-up/down navigation in /resume session selector jumps by 5 items at a time.
›Fuzzy search in /settings menu lets you type to filter settings by label.

└──▷ BREAKING ON UPGRADE

!pi.getAllTools() now returns ToolInfo[] (with name and description) instead of string[]; extensions that only need names must use .map(t => t.name).

1 more release in this issue · 2026-01-12

v0.43.0 NOTES STABLE

Pi v0.43.0 adds /scoped-models cycling, skill slash commands with fuzzy autocomplete, and new extension hooks for model changes and custom working messages.

└──▷ GET THIS VERSION

$ git clone --branch v0.43.0 https://github.com/earendil-works/pi.git
# already have the repo? check out this version:
$ git checkout v0.43.0

└──▷ TRY IT

Restrict Ctrl+P model cycling to only your preferred models for the current session, then persist the selection.

$ /scoped-models

Invoke a loaded skill directly from the command line using its registered slash command, with fuzzy matching if you forget the full name.

$ /skill:brave-search what CVEs were published this week?

Show a custom status message in the UI while your extension is doing async work during streaming.

$ ctx.ui.setWorkingMessage("Fetching threat intel...");

›New /scoped-models command lets you enable/disable models for Ctrl+P cycling; use Ctrl+S to persist choices to settings.json.
›Loaded skills are now registered as /skill:name slash commands for quick access; toggle via /settings or skills.enableSkillCommands.
›Slash command autocomplete now uses fuzzy matching (e.g., /skbra matches /skill:brave-search).
›New model_select extension hook fires on model changes (via /model, cycling, or session restore) with source and previousModel fields.
›New ctx.ui.setWorkingMessage() extension API lets extensions customize the 'Working...' message during streaming.

+5 moreshow less

›New SessionInfo.cwd field exposes a session's working directory; /resume All view now displays session cwd with loading progress.
›SessionManager.list() and SessionManager.listAll() now accept an optional onProgress callback for progress updates.
›New SessionListProgress type export for use with progress callbacks.
›/tree branch summarization now offers three options: 'No summary', 'Summarize', and 'Summarize with custom prompt'; custom prompts are appended to default summarization instructions.
›Tab in the /resume selector toggles between current-folder and all sessions.

└──▷ BREAKING ON UPGRADE

!The /branch command is renamed to /fork; RPC branch → fork and get_branch_messages → get_fork_messages; SDK branch() → fork() and getBranchMessages() → getForkMessages(); AgentSession branch() → fork() and getUserMessagesForBranching() → getUserMessagesForForking(); extension events session_before_branch → session_before_fork and session_branch → session_fork; settings value doubleEscapeAction: "branch" → "fork".
!Extension editor (ctx.ui.editor()) submit key changed from Ctrl+Enter to Enter (Shift+Enter now inserts newlines); extensions with hardcoded "ctrl+enter" hints need updating.
!SessionManager.list() and SessionManager.listAll() are now async, returning Promise<SessionInfo[]>; all callers must be updated to await them.

Was this useful?

Alibaba Qwen Code

sdk-typescript-v0.1.2 2 RELEASES · 2026-01-12 NOTES STABLE

Qwen Code SDK TypeScript v0.1.2 adds German UI, a Java SDK, stable --acp flag, multi-provider model config, and direct /approval-mode arguments.

└──▷ GET THIS VERSION

$ git clone --branch sdk-typescript-v0.1.2 https://github.com/QwenLM/qwen-code.git
# already have the repo? check out this version:
$ git checkout sdk-typescript-v0.1.2

└──▷ TRY IT

Run Qwen Code with the now-stable Agent Communication Protocol flag instead of the old experimental one.

$ qwen --acp

›Graduates --experimental-acp to the stable --acp flag for Agent Communication Protocol support.
›Adds direct argument support for the /approval-mode command, allowing approval mode to be set inline without a follow-up prompt.
›Adds multi-provider model configuration support, enabling multiple AI provider backends to be defined and selected from config.
›Adds a Java SDK (Feat/javasdk) for programmatic integration with Qwen Code.
›Adds German (de) language support via the i18n system.

+1 moreshow less

›Changes the selection order of user settings vs. workspace settings, giving workspace settings higher precedence.

└──▷ BREAKING ON UPGRADE

!The --experimental-acp flag is replaced by --acp; scripts or configs using --experimental-acp must be updated.

1 more release in this issue · 2026-01-12

v0.7.0-nightly.20260112.bde31d12 NOTES STABLE

Qwen Code v0.7.0 adds multi-provider model configuration and revised user/workspace settings selection order.

└──▷ GET THIS VERSION

$ git clone --branch v0.7.0-nightly.20260112.bde31d12 https://github.com/QwenLM/qwen-code.git
# already have the repo? check out this version:
$ git checkout v0.7.0-nightly.20260112.bde31d12

›Adds multi-provider models config support, enabling configuration of multiple AI providers within a single Qwen Code setup.

Was this useful?

◆ Local LLM Runtimes

llama.cpp

b7713 2 RELEASES · 2026-01-12 NOTES STABLE

llama.cpp b7713 adds --kv-unified flag to the batched example CLI.

└──▷ GET THIS VERSION

$ git clone --branch b7713 https://github.com/ggml-org/llama.cpp.git
# already have the repo? check out this version:
$ git checkout b7713

›Adds --kv-unified flag to the batched example CLI, enabling unified KV cache mode required for batches containing coupled sequences.

1 more release in this issue · 2026-01-12

b7712 NOTES STABLE

Vulkan memory logger now togglable via environment variable

└──▷ GET THIS VERSION

$ git clone --branch b7712 https://github.com/ggml-org/llama.cpp.git
# already have the repo? check out this version:
$ git checkout b7712

›Vulkan memory_logger is now controlled by an environment variable, allowing runtime enable/disable without recompilation.

Was this useful?

Other / Uncategorized

◆ AI OBSERVABILITY

Arize Phoenix

arize-phoenix-v12.29.0 NOTES

Phoenix v12.29.0 adds a correctness evaluator, span event attributes in the UI, and connection timeout error handling.

└──▷ GET THIS VERSION

$ git clone --branch arize-phoenix-v12.29.0 https://github.com/Arize-ai/phoenix.git
# already have the repo? check out this version:
$ git checkout arize-phoenix-v12.29.0

›Adds a correctness evaluator for assessing LLM output accuracy in evaluation workflows.
›Displays span event attributes in the Phoenix UI, making trace event data visible during span inspection.
›Adds connection timeout error handling to improve resilience when the Phoenix server is unreachable.

Was this useful?

◆ MCP TOOLING

Composio

v0.10.5 3 RELEASES · 2026-01-12 NOTES STABLE

Composio v0.10.5 adds wait-for-connections control, session-scoped execution modifiers, and auto file upload/download to the tool router.

└──▷ GET THIS VERSION

$ git clone --branch v0.10.5 https://github.com/ComposioHQ/composio.git
# already have the repo? check out this version:
$ git checkout v0.10.5

›Adds waitForConnections (TypeScript) / wait_for_connections (Python) property to manageConnections / manage_connections in toolRouter.create() / tool_router.create() to block session progression until users complete authentication.
›Adds getRawToolRouterMetaTools(sessionId, { modifySchema }) (TypeScript) / get_raw_tool_router_meta_tools(session_id, modifiers) (Python) method on the Tools class for fetching meta tools directly from a tool router session.
›Introduces SessionExecuteMetaModifiers and SessionMetaToolOptions modifier types, plus @before_execute_meta and @after_execute_meta decorators (Python), exposing sessionId / session_id to beforeExecute / afterExecute hooks for session-scoped tool execution control.
›Adds auto_upload_download_files boolean flag to the Composio() constructor (Python) to control automatic file upload and download behaviour.
›Upgrades the Vercel provider (TypeScript) to support Vercel AI SDK v6.

2 more releases in this issue · 2026-01-12

@composio/[email protected] NOTES STABLE

Composio Vercel SDK 0.4.0 adds dedicated tools support for the tool router.

└──▷ 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]

›Adds support for dedicated tools for the tool router in @composio/vercel.

v0.10.4 NOTES STABLE

Composio v0.10.4 adds Anthropic Claude Code Agents provider, LangChain v1 port for TypeScript, and openWorldHint tag filters for the tool router.

└──▷ GET THIS VERSION

$ git clone --branch v0.10.4 https://github.com/ComposioHQ/composio.git
# already have the repo? check out this version:
$ git checkout v0.10.4

›Adds openWorldHint as a tag filter option in the tool router, enabling agents to distinguish open-world from closed-world tool scopes.
›Adds Anthropic Claude Code Agents as a new provider in the core SDK.
›Ports the LangChain integration to LangChain v1 in the TypeScript SDK.
›Adds support for enable/disable tags and search within toolkits.

Was this useful?

// END OF TRANSMISSION

The AI Toolchain · curated daily · January 12, 2026
Releases & features from the open-source and commercial AI tools that practitioners actually run.

The AI Toolchain — issue -217, January 12, 2026