The AI Toolchain — issue -317, October 3, 2025

Agno v2.1.1 adds a prefix option for AGUI, WhatsApp, and Slack interfaces to support multiple interfaces on one AgentOS instance.

$ git clone --branch v2.1.1 https://github.com/agno-agi/agno.git
# already have the repo? check out this version:
$ git checkout v2.1.1

• ›Adds a prefix parameter to AGUI, Whatsapp, and Slack interfaces, enabling multiple interfaces to run on the same AgentOS instance with distinct route prefixes.

LlamaIndex v0.14.4 adds Bedrock AgentCore Memory, Apache Solr vector store, structured outputs for OpenAILike, and Claude Sonnet 4.5 support.

$ git clone --branch v0.14.4 https://github.com/run-llama/llama_index.git
# already have the repo? check out this version:
$ git checkout v0.14.4

• ›Adds llama-index-memory-bedrock-agentcore (v0.1.0) with a new Bedrock AgentCore Memory integration for persistent agent memory backed by AWS.
• ›Adds llama-index-vector-stores-solr (v0.1.0) with a new ApacheSolrVectorStore integration for using Apache Solr as a vector store backend.
• ›Adds structured outputs support to OpenAILike in llama-index-llms-openai-like and llama-index-llms-openai.
• ›Adds support for Claude Sonnet 4.5 (anthropic-sonnet-4-5) in llama-index-llms-anthropic.
• ›Adds support for Claude Sonnet 4.5 in llama-index-llms-bedrock-converse.

PydanticAI v1.0.15 adds image generation support, streaming events API, and new ModelResponse convenience methods.

$ git clone --branch v1.0.15 https://github.com/pydantic/pydantic-ai.git
# already have the repo? check out this version:
$ git checkout v1.0.15

async for event in agent.run_stream_events('Summarize this document', deps=deps):
    print(event)

result = await agent.run('What is 2 + 2?')
print(result.response.text)

• ›Adds AgentRunResult.response convenience method to retrieve the latest model response from a completed agent run.
• ›Adds ModelResponse.text, ModelResponse.thinking, ModelResponse.files, ModelResponse.images, ModelResponse.tool_calls, and ModelResponse.builtin_tool_calls convenience methods for accessing parts of a model response.
• ›Adds Agent.run_stream_events() convenience method as a shorthand wrapper around run(event_stream_handler=...).
• ›Supports image generation and image output with Google and OpenAI providers.
• ›Adds content (e.g. files) returned by a tool to FunctionToolResultEvent, making tool output accessible in event streams.

GitHub Copilot CLI v0.0.334 adds paste compaction and a context-limit warning to help manage long sessions.

$ git clone --branch v0.0.334 https://github.com/github/copilot-cli.git
# already have the repo? check out this version:
$ git checkout v0.0.334

• ›Pasting more than 10 lines now displays a compact token (e.g. [Paste #1 - 15 lines]) instead of flooding the terminal.
• ›Adds a warning when conversation context reaches ≤20% of the model's remaining limit, prompting users to start a new session before truncation occurs.
• ›Adds the current version to startup logs to aid in bug reporting.

Lightweight coding agent that runs in your terminal

Codex CLI v0.44.0 adds cloud task management, streamable HTTP MCP servers, custom prompt arguments, and a refreshed TUI.

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

$ codex cloud

• ›New TUI refresh: updated chat composer, bottom pane, two-stage model picker, and upgrade nudge.
• ›Custom prompts now support named and positional (numeric) arguments via the /prompts: prefix.
• ›Adds support for streamable HTTP MCP servers.
• ›New codex cloud subcommand (very experimental) for managing cloud tasks.
• ›Experimental reduction of repetitive approval prompts for read-only (safe) commands on Windows via PowerShell command safety classification.

LocalAI v3.6.0 adds L4T device support for Kokoro, multilingual Chatterbox TTS, and six new gallery models including IBM Granite 4.0 and Qwen image editing.

$ git clone --branch v3.6.0 https://github.com/mudler/LocalAI.git
# already have the repo? check out this version:
$ git checkout v3.6.0

• ›Adds Kokoro TTS backend support for L4T (Linux for Tegra) devices, enabling on-device speech synthesis on NVIDIA Jetson hardware.
• ›Adds multilingual support to the Chatterbox TTS backend.
• ›Adds qwen-image-edit-2509 to the model gallery for image editing workflows.
• ›Adds whisper-turbo via the whisper.cpp backend to the model gallery.
• ›Adds ibm-granite_granite-4.0-h-small, ibm-granite_granite-4.0-h-tiny, ibm-granite_granite-4.0-h-micro, and ibm-granite_granite-4.0-micro to the model gallery.

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

Ollama v0.12.4 enables flash attention by default for Qwen3 models and adds a new override to disable it.

$ git clone --branch v0.12.4 https://github.com/ollama/ollama.git
# already have the repo? check out this version:
$ git checkout v0.12.4

$ OLLAMA_FLASH_ATTENTION=0 ollama run qwen3

• ›Enables flash attention by default for Qwen3 and Qwen3 Coder models, improving inference performance without manual configuration.
• ›Adds support for overriding default flash attention via OLLAMA_FLASH_ATTENTION=0 for models where it is enabled by default.

Phoenix 12.2.0 adds root_path to launch_app, prompt version editing and tag loading in playground, and dataset label bulk management.

$ git clone --branch arize-phoenix-v12.2.0 https://github.com/Arize-ai/phoenix.git
# already have the repo? check out this version:
$ git checkout arize-phoenix-v12.2.0

import phoenix as px
px.launch_app(root_path="/phoenix")

• ›Adds root_path parameter to launch_app to control the URL prefix when serving Phoenix behind a reverse proxy or subpath.
• ›Enables bulk add/remove of labels on dataset examples via the UI.
• ›Adds dataset label color validation to enforce valid label colors.
• ›Supports editing prompt versions directly in the playground.
• ›Supports loading a prompt into the playground by tag.

Composio v0.8.20 adds allow_multiple for connected accounts and OpenAI Responses API support in the Python SDK.

$ git clone --branch v0.8.20 https://github.com/ComposioHQ/composio.git
# already have the repo? check out this version:
$ git checkout v0.8.20

• ›Adds allow_multiple option to the Python SDK when creating connected accounts, enabling multiple connections to the same integration.
• ›Adds support for the OpenAI Python Responses API in the Composio Python SDK.