The AI Toolchain — issue -311, October 10, 2025

Agno v2.1.4 adds workflow history, a GoogleDriveTools class, AG-UI custom events, and tool post-hooks on failure.

$ git clone --branch v2.1.4 https://github.com/agno-agi/agno.git
# already have the repo? check out this version:
$ git checkout v2.1.4

• ›Adds GoogleDriveTools class to give agents read and write access to Google Drive.
• ›Adds workflow history support, enabling continuous conversational context across all or individual workflow steps.
• ›Extends tool post-hooks to also execute when a tool run fails with an exception, enabling failure-specific cleanup or logging logic.
• ›AG-UI integration now delivers Agno custom events to the AG-UI interface in the standard AG-UI custom event format.
• ›Parallel step event streaming in workflows now yields events immediately as they are produced rather than collecting all events and yielding at the end.

langchain-anthropic 1.0.0a4 adds web fetch beta, code execution, MCP connector, files API, web search, citations streaming, cache_control kwarg, and more.

$ git clone --branch langchain-anthropic==1.0.0a4 https://github.com/langchain-ai/langchain.git
# already have the repo? check out this version:
$ git checkout langchain-anthropic==1.0.0a4

from langchain_anthropic import ChatAnthropic
from langchain_core.tools import tool

@tool
def get_weather(city: str) -> str:
    """Get weather for a city."""
    return f"Sunny in {city}"

model = ChatAnthropic(model="claude-opus-4-5").bind_tools(
    [get_weather],
    parallel_tool_calls=True
)
response = model.invoke("What is the weather in Paris and London?")

• ›Adds cache_control as a kwarg to ChatAnthropic for fine-grained prompt caching control.
• ›Adds parallel_tool_calls support to ChatAnthropic.
• ›Adds support for built-in tools in ChatAnthropic.
• ›Adds web fetch beta capability to ChatAnthropic for fetching web content during inference.
• ›Adds web search support to ChatAnthropic.

$ git clone --branch langchain==1.0.0a13 https://github.com/langchain-ai/langchain.git
# already have the repo? check out this version:
$ git checkout langchain==1.0.0a13

from langchain_v1 import ToolCallLimitMiddleware, create_agent

agent = create_agent(
    model=model,
    tools=[search, calculator],
    middleware=[ToolCallLimitMiddleware(max_tool_calls=5)],
)

import asyncio
from langchain_v1 import create_agent

agent = create_agent(model=model, tools=[search])
result = await agent.ainvoke({'messages': [{'role': 'user', 'content': 'What is the weather in Paris?'}]})

GitHub Copilot CLI v0.0.339 improves MCP server setup with full shell-style command input in /mcp add

$ git clone --branch v0.0.339 https://github.com/github/copilot-cli.git
# already have the repo? check out this version:
$ git checkout v0.0.339

• ›The Command field in /mcp add now accepts a full shell-style command string to start an MCP server, replacing the previous comma-separated argument syntax.

The open source coding agent.

OpenCode v0.14.7 adds Vertex AI, image reading, session forking, SSE streaming, and a new output-format flag

$ git clone --branch v0.14.7 https://github.com/sst/opencode.git
# already have the repo? check out this version:
$ git checkout v0.14.7

$ opencode debug config

import { createOpencodeTui } from 'opencode'

await createOpencodeTui()

• ›Adds Vertex AI support via google-vertex and google-vertex-anthropic providers
• ›Enables the read tool to handle images, not just text files
• ›New --output-format flag streams JSON output from commands
• ›New debug config command for inspecting resolved configuration
• ›Adds session forking functionality and corresponding API endpoints

OpenHands: AI-Driven Development

OpenHands 0.59.0 adds Lemonade as a supported LLM provider.

$ git clone --branch 0.59.0 https://github.com/All-Hands-AI/OpenHands.git
# already have the repo? check out this version:
$ git checkout 0.59.0

• ›Adds Lemonade Provider as a supported LLM backend option.

$ git clone --branch 1.0.0-cli https://github.com/All-Hands-AI/OpenHands.git
# already have the repo? check out this version:
$ git checkout 1.0.0-cli

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

gemini-cli v0.10.0-nightly adds generalized path correction, failed-response retry, diff stats in telemetry, and a full-width display setting.

$ git clone --branch v0.10.0-nightly.20251010.558be873 https://github.com/google-gemini/gemini-cli.git
# already have the repo? check out this version:
$ git checkout v0.10.0-nightly.20251010.558be873

• ›Generalizes path correction logic across all tools, improving reliability when referencing files in any tool context.
• ›Adds automatic retry of failed model responses via an extra prompt, reducing dead-end interactions.
• ›Adds diff stats to tool call metrics in telemetry for richer observability of file-editing operations.
• ›Adds a full-width display setting for narrow screen layouts.

Get up and running with Kimi-K2.6, GLM-5.2, MiniMax, DeepSeek, gpt-oss, Qwen, Gemma and other models.

Ollama v0.12.5 adds structured output support for thinking models via /api/chat

$ git clone --branch v0.12.5 https://github.com/ollama/ollama.git
# already have the repo? check out this version:
$ git checkout v0.12.5

$ curl http://localhost:11434/api/chat -d '{"model": "deepseek-r1", "messages": [{"role": "user", "content": "Extract the name and age from: John is 30 years old."}], "format": {"type": "object", "properties": {"name": {"type": "string"}, "age": {"type": "integer"}}, "required": ["name", "age"]}, "stream": false}'

• ›Supports structured outputs for thinking models when using the /api/chat API

v3.14 adds /v1/internal/logits endpoint for exllamav3 loaders and qwen3-next model support via fla.

$ git clone --branch v3.14 https://github.com/oobabooga/textgen.git
# already have the repo? check out this version:
$ git checkout v3.14

• ›Implements the /v1/internal/logits API endpoint for the exllamav3 and exllamav3_hf loaders.
• ›Adds fla to requirements for Exllamav3 to enable support for qwen3-next models.
• ›Improves handling of multi-GPU setups when using Transformers with bitsandbytes (load-in-8bit and load-in-4bit).

Phoenix v12.5.0 adds a viewer role, dataset split selection, example table filtering, and annotation sorting in sessions.

$ git clone --branch arize-phoenix-v12.5.0 https://github.com/Arize-ai/phoenix.git
# already have the repo? check out this version:
$ git checkout arize-phoenix-v12.5.0

• ›Adds a viewer role for read-only access control in Phoenix.
• ›Enables selection of dataset splits alongside datasets in the UI.
• ›Adds filtering to the examples table and a split management filter menu.
• ›Enables sorting on annotations in the sessions table.
• ›Adds editable labels on datasets.

Langfuse v3.117.0 adds named prompt experiment runs, self-serve account deletion, and improved Docker Compose flexibility for self-hosting.

$ git clone --branch v3.117.0 https://github.com/langfuse/langfuse.git
# already have the repo? check out this version:
$ git checkout v3.117.0

• ›Adds self-serve account deletion and display name update under account settings.
• ›Allows naming runs in prompt experiments for easier identification and comparison.
• ›Enhances Docker Compose configuration flexibility for self-hosting platforms.

Weave v0.52.9 adds OpenAI Realtime support, image tracing, OTEL user ID parsing, custom call-page content, and a new object query filter.

$ git clone --branch v0.52.9 https://github.com/wandb/weave.git
# already have the repo? check out this version:
$ git checkout v0.52.9

import weave

client = weave.init('my-project')
result = client.server.objs_query(
    weave.trace_server.trace_server_interface.ObjQueryReq(
        project_id='my-entity/my-project',
        filter=weave.trace_server.trace_server_interface.ObjectVersionFilter(
            exclude_base_object_classes=['Model', 'Dataset']
        )
    )
)

• ›Adds exclude_base_object_classes to the objects query filter, letting callers exclude base object classes when querying the object store.
• ›Supports OpenAI Realtime API tracing — conversations over the realtime websocket interface are now captured as Weave traces.
• ›Adds image support to the trace server so image data can be stored and retrieved as part of call inputs/outputs.
• ›Parses user ID from OTEL spans, surfacing per-user attribution in OpenTelemetry-sourced traces.
• ›Enables user code to define custom content displayed on a call's detail page.

Milvus 2.6.3 adds manual L0 compaction, gRPC tokenizer, sparse filters, and new autoindex options for int8 vectors.

$ git clone --branch v2.6.3 https://github.com/milvus-io/milvus.git
# already have the repo? check out this version:
$ git checkout v2.6.3

• ›Adds new configuration options for the int8 vector type in autoindexing.
• ›Adds parameter items to control hybrid search requery policy.
• ›Enables autoid-enabled collections to accept explicit primary key values on insert, including via the Go SDK.
• ›Adds manual compaction support for L0 segments.
• ›Integrates a gRPC tokenizer for enhanced query flexibility.

Weaviate v1.32.11 adds image support in generative-cohere, new debug endpoints, and a broad set of new observability metrics.

$ git clone --branch v1.32.11 https://github.com/weaviate/weaviate.git
# already have the repo? check out this version:
$ git checkout v1.32.11

• ›Adds debug endpoints for shard and lock status monitoring via feat(debug) additions.
• ›Adds new compaction metrics for LSM storage observability.
• ›Adds async replication metrics for tracking replication health.
• ›Adds LSM WAL recovery metrics.
• ›Adds memtable flushing metrics.