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.
Agno v2.0.6 adds FileGenerationTools for PDF/CSV/JSON/TXT output and a new Nexus Router Model class.
└──▷ GET THIS VERSION
$ git clone --branch v2.0.6 https://github.com/agno-agi/agno.git
# already have the repo? check out this version:$ git checkout v2.0.6
└──▷ USE IT
Give an agent the ability to produce downloadable file artifacts (PDF, CSV, JSON, TXT) as part of its responses.
python
from agno.agent import Agent
from agno.tools.file_generation import FileGenerationTools
agent = Agent(
tools=[FileGenerationTools()],
description="An agent that can generate and save file artifacts.",
)
agent.print_response("Summarise this dataset and save it as a CSV report.")
›Adds FileGenerationTools toolkit, enabling agents to generate file artifacts in PDF, CSV, JSON, and TXT formats.
›Adds a Model class for the Nexus Router, exposing it as a first-class model provider.
›Stores the main workflow input in WorkflowRunOutput and persists it to the database.
langchain-core 1.0.0a4 adds standard AWS/Bedrock content blocks, public is_openai_data_block filtering, custom Mermaid URLs, and more.
└──▷ GET THIS VERSION
$ git clone --branch langchain-core==1.0.0a4 https://github.com/langchain-ai/langchain.git
# already have the repo? check out this version:$ git checkout langchain-core==1.0.0a4
└──▷ USE IT
Filter a list of content blocks to keep only OpenAI data blocks using the now-public is_openai_data_block.
python
from langchain_core.messages.content import is_openai_data_block
blocks = message.content
data_blocks = [b for b in blocks if is_openai_data_block(b)]
Strip PostgreSQL NUL bytes from LLM output before inserting into a Postgres store to avoid DataError.
python
from langchain_core.utils import sanitize_for_postgres
clean_text = sanitize_for_postgres(llm_output)
cursor.execute('INSERT INTO results (content) VALUES (%s)', (clean_text,))
›Makes is_openai_data_block public and adds filtering support for OpenAI data blocks.
›Adds id field to Document objects passed to the filter callback in InMemoryVectorStore similarity search.
›Adds web_search to the list of recognized OpenAI built-in tools.
›Supports AWS Bedrock document content blocks in msg_content_output.
›Supports standard AWS content blocks (v1 standard content for AWS).
+14 moreshow less
›Allows overriding ls_model_name from kwargs at call time.
›Allows custom Mermaid diagram URL via the new custom Mermaid URL feature.
›Supports adding PromptTemplates with formats other than f-string.
›Adds sanitize_for_postgres utility to strip PostgreSQL NUL bytes and prevent DataError.
›Adds an option to make deserialization more permissive.
›Zeros out token costs for cache hits in token usage tracking.
›Adds image_generation to the list of known OpenAI tools.
›Exposes tool message recognized block types.
›Adds additional hashing options to the indexing API and warns on SHA-1 usage.
›Traces response body on error for improved observability.
›Supports Union type args in strict mode of OpenAI function calling and structured output.
langchain-mistralai 0.2.12 allows overriding ls_model_name at call time via kwargs.
└──▷ GET THIS VERSION
$ git clone --branch langchain-mistralai==0.2.12 https://github.com/langchain-ai/langchain.git
# already have the repo? check out this version:$ git checkout langchain-mistralai==0.2.12
›Supports overriding ls_model_name via kwargs at invocation time, enabling per-call model name labeling in LangSmith tracing.
LangChain Core 1.0.0a3 adds standard content blocks, new OpenAI tools, AWS Bedrock document support, and multiple tracing and filtering improvements.
└──▷ GET THIS VERSION
$ git clone --branch langchain-core==1.0.0a3 https://github.com/langchain-ai/langchain.git
# already have the repo? check out this version:$ git checkout langchain-core==1.0.0a3
└──▷ USE IT
Filter data content blocks using the now-public is_openai_data_block to pre-process message content before sending to a model.
python
from langchain_core.messages.content import is_openai_data_block
blocks = message.content
data_blocks = [b for b in blocks if is_openai_data_block(b)]
Use sanitize_for_postgres to strip NUL bytes from text before storing documents in a PostgreSQL-backed vector store.
python
from langchain_core.utils import sanitize_for_postgres
clean_text = sanitize_for_postgres(raw_document_text)
›Makes is_openai_data_block public and adds filtering support for data content blocks.
›Adds id field to Document objects passed to the filter callback in InMemoryVectorStore similarity search.
›Adds web_search to the OpenAI tools list recognized by LangChain Core.
›Supports AWS Bedrock document content blocks in msg_content_output.
›Supports adding PromptTemplates with formats other than f-string.
+13 moreshow less
›Allows overriding ls_model_name from kwargs during tracing.
›Allows specifying a custom Mermaid diagram URL via the new custom Mermaid URL feature.
›Adds sanitize_for_postgres utility to strip PostgreSQL NUL bytes that cause DataError.
›Adds an option to make deserialization more permissive.
›Introduces standard content blocks, IDs, translators, and normalization as a major new content-handling framework.
›Autogenerates filenames when converting file content blocks to OpenAI format.
›Zeros out token costs for cache hits in token usage tracking.
$ git clone --branch v0.3.1 https://github.com/openai/openai-agents-python.git
# already have the repo? check out this version:$ git checkout v0.3.1
└──▷ USE IT
Attach a plain-English description to a function tool parameter without a separate docstring, using Annotated.
python
from typing import Annotated
from agents import function_tool
@function_tool
def search(query: Annotated[str, 'The search query to look up'], max_results: Annotated[int, 'Maximum number of results to return'] = 10) -> list[str]:
...
›Supports typing.Annotated types for function tool parameter descriptions, letting developers embed param metadata directly in type hints.
›Adds more options to Agent#as_tool for finer control when exposing an agent as a callable tool.
›Exports user_agent_override context manager for overriding the HTTP User-Agent header at runtime.
›Adds input audio noise reduction for realtime voice sessions via the Realtime API.
›Migrates STT streaming to match the GA Realtime API.
+3 moreshow less
›Adds session encryption support using the cryptography library in the Sessions implementation.
›Supports Anthropic extended thinking and interleaved thinking in agent runs.
›Adds a warning when agent names transform into conflicting function names.
└──▷ BREAKING ON UPGRADE
!Voice STT streaming has been migrated to match the GA Realtime API — existing STT streaming integrations may need to be updated.
PydanticAI v1.0.9 adds RunContext retry introspection and streams built-in tool calls from OpenAI, Google, and Anthropic.
└──▷ GET THIS VERSION
$ git clone --branch v1.0.9 https://github.com/pydantic/pydantic-ai.git
# already have the repo? check out this version:$ git checkout v1.0.9
└──▷ USE IT
Gate expensive fallback logic so it only runs on the last allowed attempt inside a tool.
python
from pydantic_ai import RunContext
async def my_tool(ctx: RunContext[None], query: str) -> str:
if ctx.last_attempt:
return f'Final attempt reached (max={ctx.max_retries}), returning cached result'
result = call_external_api(query)
return result
›Adds RunContext.max_retries and RunContext.last_attempt so tool functions can inspect retry limits and detect whether the current invocation is the final attempt.
›Streams built-in tool calls from OpenAI, Google, and Anthropic and returns them on the next request, enabling support for OpenAI reasoning models.
›Includes built-in tool calls and their results in OpenTelemetry (OTel) messages for full observability of tool interactions.
Continue config-yaml 1.19.0 adds an invokable option for markdown rules and more robust context length validation.
└──▷ GET THIS VERSION
$ git clone --branch @continuedev/[email protected] https://github.com/continuedev/continue.git
# already have the repo? check out this version:$ git checkout @continuedev/[email protected]
›Adds invokable option to markdown rule creation, allowing rules to be explicitly triggered rather than applied automatically.
›Adds more robust context length validation to catch configuration issues before they cause runtime failures.
›Uses recommended-models function for tools, improving default model selection when configuring tool use.
Continue v1.2.4 adds CLI shell mode, terminal security warnings, dynamic tool policies, and a new experimental range tool for file operations.
└──▷ GET THIS VERSION
$ git clone --branch v1.2.4-vscode https://github.com/continuedev/continue.git
# already have the repo? check out this version:$ git checkout v1.2.4-vscode
└──▷ TRY IT
Run the CLI against a specific config file when connecting to a remote agent, without modifying your default config.
$ cn remote --config ./my-project-config.yaml
Bootstrap a new project context from the CLI using the /init slash command.
$ /init
Check CLI diagnostics — environment, model, and config details — from inside a running session.
$ /info
›Adds --config flag to cn remote to pass a config file path at invocation time.
›Adds /init slash command to the CLI for project initialization.
›Adds /info screen diagnostics to the CLI, surfacing environment and configuration details.
›Adds CLI shell mode, enabling an interactive shell session via the Continue CLI.
›Adds ctrl+c once to clear chat input and twice to exit in the CLI TUI.
+11 moreshow less
›Adds terminal command security warnings to both the VS Code extension and CLI, flagging potentially dangerous commands before execution.
›Introduces dynamic tool policies, with the terminal tool as the first implementation, allowing per-tool execution policy control.
›Adds an experimental range tool for scoped file read/write operations.
›Adds filepath as a variable available in apply prompt templates.
›Adds org-level policies for indexing and MCP (Model Context Protocol) configuration.
›Runs dependent indexers automatically based on the active context provider.
›Returns token usage data for OpenAI providers.
›Shows the current git branch in the CLI interface when available.
›Adds a commit signature feature to the CLI.
›Skips telemetry in the CLI when no internet connection is detected.
›Updates supported models for the Scaleway provider.
Qwen Code v0.0.12-nightly.1 adds skipLoopDetection config option and a confirmation prompt to /init.
└──▷ GET THIS VERSION
$ git clone --branch v0.0.12-nightly.1 https://github.com/QwenLM/qwen-code.git
# already have the repo? check out this version:$ git checkout v0.0.12-nightly.1
›Adds skipLoopDetection configuration option to disable loop detection during agentic runs.
›Enhances /init command with a confirmation prompt before executing initialization.
An open-source AI agent that brings the power of Gemini directly into your terminal.
gemini-cli v0.7.0-nightly adds JSON output, /chat share, Ctrl+R history search, ripgrep by default, interactive shell, and undo/redo.
└──▷ GET THIS VERSION
$ git clone --branch v0.7.0-nightly.20250918.2722473a https://github.com/google-gemini/gemini-cli.git
# already have the repo? check out this version:$ git checkout v0.7.0-nightly.20250918.2722473a
└──▷ TRY IT
Search your session history to re-run a previous prompt without retyping it.
$ # Inside an active gemini-cli session, press Ctrl+R then type keywords to fuzzy-search prior commands
Update all installed gemini-cli extensions to their latest versions.
$ gemini extensions update
›Adds --output json flag for machine-readable JSON output format.
›New /chat share slash command to share the current chat session.
›Adds Ctrl+R command history search in the input prompt.
›Adds Ctrl+Z / Ctrl+R undo/redo shortcuts in the input editor.
›Enables ripgrep by default for faster file search, with auto-redownload if the binary is deleted.
+15 moreshow less
›Supports interactive commands via a virtual terminal in the shell tool.
›New extensions update command to update installed extensions.
›Supports installing a git extension at a specific ref and checking for git extension updates.
›Adds .geminiignore support to the glob tool to exclude files from searches.
›Auto-approves pending tool calls when auto_edit or yolo mode is activated.
›Adds a confirmation dialog for disabling loop detection for the current session.
›Caps shell output truncation threshold to the remaining context window size.
›Adds permission prompt when installing a local extension that includes MCP servers.
›Adds local/remote designation for MCP servers in extensions.
›Adds auth token validation and support to the VS Code IDE companion server.
›Introduces classifier-based model routing strategy and useModelRouter feature flag.
›Introduces hybrid token storage when the corresponding security flag is enabled.
›Introduces encrypted OAuth token storage flag.
›Checks for IDE diffing capabilities before opening diffs, with an improved diff communication protocol.
›IDE companion extension specification added, defining the diffing interface.
$ git clone --branch arize-phoenix-client-v1.19.0 https://github.com/Arize-ai/phoenix.git
# already have the repo? check out this version:$ git checkout arize-phoenix-client-v1.19.0
›Adds Experiments↔Evals 2.0 compatibility, enabling experiments to work with the Evals 2.0 evaluation format.
LanceDB python-v0.25.1-beta.2 adds target_partition_size parameter for index tuning.
└──▷ GET THIS VERSION
$ git clone --branch python-v0.25.1-beta.2 https://github.com/lancedb/lancedb.git
# already have the repo? check out this version:$ git checkout python-v0.25.1-beta.2
›Adds target_partition_size parameter to control partition sizing during index creation.
LanceDB v0.22.1-beta.2 adds target_partition_size parameter for index tuning.
└──▷ GET THIS VERSION
$ git clone --branch v0.22.1-beta.2 https://github.com/lancedb/lancedb.git
# already have the repo? check out this version:$ git checkout v0.22.1-beta.2
›Adds target_partition_size parameter to control partition sizing during index creation.
THE AI TOOLCHAIN