Heads up This site is currently under heavy development.
Subscribe Get it delivered — the daily firehose, filtered to the tools you run, plus the documentation changes vendors never announce. Compare plans →

The AI Toolchain — issue -237, December 23, 2025

THE AI TOOLCHAIN NO. -237
Tail
THE DAILY RELEASE FIREHOSE
PUBLISHED DECEMBER 23, 2025 · EVERY WEEKDAY
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   # 8 tools matched
AI & LLM Tooling
◆  AI Agent Frameworks

Agno (formerly Phidata)

Sources Release notes → v2.3.21 NOTES

Agno v2.3.21 brings AgentAsJudge evals to AgentOS with full run configuration and listing support.

└──▷ GET THIS VERSION
$ git clone --branch v2.3.21 https://github.com/agno-agi/agno.git
# already have the repo? check out this version:
$ git checkout v2.3.21
  • AgentAsJudge evaluations are now fully supported on AgentOS: configure and trigger new runs, and view existing runs alongside other evals on the Evals page.
Was this useful?

LangChain

Sources Release notes → langchain-classic==1.0.1 NOTES

LangChain Classic 1.0.1 adds google_genai embedding support, extras on BaseTool, and effort support in Anthropic.

└──▷ GET THIS VERSION
$ git clone --branch langchain-classic==1.0.1 https://github.com/langchain-ai/langchain.git
# already have the repo? check out this version:
$ git checkout langchain-classic==1.0.1
└──▷ USE IT
Initialize a Google Generative AI embedding model using the unified init_embeddings interface.
python
from langchain.embeddings import init_embeddings

embeddings = init_embeddings(model="text-embedding-004", provider="google_genai")
Pass provider-specific metadata through a tool definition using the new extras field on BaseTool.
python
from langchain_core.tools import BaseTool

class MyTool(BaseTool):
    name: str = "my_tool"
    description: str = "Does something useful"
    extras: dict = {"cache_control": {"type": "ephemeral"}}

    def _run(self, query: str) -> str:
        return query
  • Adds google_genai provider support to init_embeddings, enabling Google Generative AI embedding models via the standard initializer.
  • Adds extras field on BaseTool (in core and anthropic) for passing arbitrary provider-specific metadata through tool definitions.
  • Adds effort parameter support to the Anthropic integration for controlling model reasoning effort.
  • Enhances init_chat_model with improved validation to catch misconfigured model/provider combinations earlier.
  • Switches run IDs to UUID v7, providing time-ordered identifiers for LangChain runs.
Was this useful?

PydanticAI

Sources Release notes → v1.38.0 NOTES

PydanticAI v1.38.0 adds local timestamps to request/response models and typed RunContext support in TextOutput signatures.

└──▷ GET THIS VERSION
$ git clone --branch v1.38.0 https://github.com/pydantic/pydantic-ai.git
# already have the repo? check out this version:
$ git checkout v1.38.0
  • Adds typed RunContext[Deps] support in TextOutput function signatures, enabling dependency-injected output handlers.
  • Adds local timestamps to request and response models, with provider timestamps surfaced in provider_details.
  • Supports VideoUrl.vendor_metadata for GCS URIs on the Google Vertex provider.
Was this useful?
◆  AI Coding Agents

SST OpenCode

Sources Release notes → v1.0.191 NOTES

The open source coding agent.

OpenCode v1.0.191 adds MCP OAuth debugging, per-agent skill filtering, and compaction hook prompt replacement.

└──▷ GET THIS VERSION
$ git clone --branch v1.0.191 https://github.com/sst/opencode.git
# already have the repo? check out this version:
$ git checkout v1.0.191
  • Adds ability to debug MCP OAuth via the mcp CLI subcommand.
  • Adds per-agent filtering to skill tool descriptions, letting agents see only relevant tools.
  • Enables compaction hooks to replace the prompt entirely, giving plugins full control over context compaction.
Was this useful?

Earendil Works Pi

Sources Release notes → v0.27.2 NOTES

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

Hooks can now skip conversation restore on branch for checkpoint-style session management.

└──▷ GET THIS VERSION
$ git clone --branch v0.27.2 https://github.com/earendil-works/pi.git
# already have the repo? check out this version:
$ git checkout v0.27.2
  • Adds skipConversationRestore: true return value for before_branch hooks, allowing branched sessions to be created without restoring conversation messages.
Was this useful?
◆  Local LLM Runtimes

llama.cpp

Sources Release notes → b7524 2 RELEASES · 2025-12-23 NOTES STABLE

llama.cpp server now reports 0% processing state via return_progress

└──▷ GET THIS VERSION
$ git clone --branch b7524 https://github.com/ggml-org/llama.cpp.git
# already have the repo? check out this version:
$ git checkout b7524
  • Extends return_progress in the llama.cpp server to also report the 0% processing state, giving clients visibility into the earliest stage of request handling.
1 more release in this issue · 2025-12-23
b7515 NOTES STABLE

llama.cpp b7515 adds Granite Embedding and ModernBERT model support with HuggingFace-to-GGUF conversion.

└──▷ GET THIS VERSION
$ git clone --branch b7515 https://github.com/ggml-org/llama.cpp.git
# already have the repo? check out this version:
$ git checkout b7515
  • Adds support for Granite Embedding models, including HuggingFace-to-GGUF conversion via convert_hf_to_gguf.py and inference via llama-embedding.
  • Adds ModernBERT architecture support (encoder-only, without head.norm) with alternating local/global sliding window attention, GEGLU activation, and alternating RoPE, implemented in src/models/modern-bert.cpp.
  • Adds swa_period (sliding window attention period) GGUF hyperparameter, written and read via gguf-py/gguf/gguf_writer.py and src/llama-hparams.h, enabling dense-every-N-layers SWA patterns for ModernBERT-family models.
Was this useful?
Other / Uncategorized
◆  MCP TOOLING

Composio

Sources Release notes → v0.10.2 NOTES

Composio v0.10.2 adds Anthropic Claude Code Agents provider, LangChain v1 support, and openWorldHint tool router filtering.

└──▷ GET THIS VERSION
$ git clone --branch v0.10.2 https://github.com/ComposioHQ/composio.git
# already have the repo? check out this version:
$ git checkout v0.10.2
  • Adds openWorldHint as a tag filter for the tool router, enabling finer-grained control over which tools are surfaced by the router.
  • Adds Anthropic Claude Code Agents as a new provider in the core integration layer.
  • Ports the TypeScript LangChain integration to LangChain v1.
Was this useful?
◆  AI OBSERVABILITY

Langfuse

Sources Release notes → v3.143.0 NOTES

Langfuse v3.143.0 adds tool-call observation filtering, org details in the project API, and a virtualized JSON viewer with search and media.

└──▷ GET THIS VERSION
$ git clone --branch v3.143.0 https://github.com/langfuse/langfuse.git
# already have the repo? check out this version:
$ git checkout v3.143.0
  • Adds org details to the project API endpoint, exposing organization-level context alongside project data.
  • Enables filtering observations by tool calls in the tracing UI, letting practitioners narrow traces to specific tool invocations.
  • Introduces a fully virtualized JSON Beta view with search and media attachment support in the advanced JSON viewer.
Was this useful?
my-toolchain — 0 tools
paste an install list to detect your tools

A brew list, a Brewfile, requirements.txt, a Dockerfile — or just the product names, free-form. Nothing leaves your browser.

    browse all tools →