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 v2.4.0 adds KnowledgeProtocol, Agent Builder persistence, new lifecycle events, and GCS file inputs for Gemini
└──▷ GET THIS VERSION
$ git clone --branch v2.4.0 https://github.com/agno-agi/agno.git
# already have the repo? check out this version:$ git checkout v2.4.0
└──▷ USE IT
Point TavilyTools at a self-hosted Tavily endpoint instead of the public API.
python
from agno.tools.tavily import TavilyTools
tools = TavilyTools(api_base_url="https://tavily.internal.example.com")
Restore XML-tagged instructions for an agent that relies on structured <instructions> blocks in its system prompt.
python
from agno.agent import Agent
agent = Agent(
instructions=["Always respond in bullet points."],
add_instruction_tags=True
)
›Adds api_base_url parameter to TavilyTools to point at custom-hosted Tavily instances.
›Adds add_instruction_tags=True option on Agent and Team to restore wrapping instructions in <instructions> XML tags (now omitted by default).
›Introduces KnowledgeProtocol interface so any custom knowledge implementation can be used with Agent and Team (only the main Agno implementation supports AgentOS Knowledge management).
›Introduces Agent Builder: Agent, Team, and Workflow configurations can now be persisted and managed in a database via new AgentOS endpoints for programmatic creation, retrieval, and updates.
›Adds new lifecycle events: ModelRequestStarted, ModelRequestCompleted, CompressionStarted, and CompressionCompleted; updates MemoryUpdateCompleted to include memory content.
+5 moreshow less
›Adds direct GCS URI and external URL support for Gemini file inputs.
›Adds db parameter to the AgentOS class that propagates to all agents, teams, and workflows without a database set, and also serves as the tracing database.
›Introduces update_memory_on_run as the replacement for the deprecated enable_user_memories.
›Replaces DDG web search tool with a generic WebSearchTools as the new default for web search in cookbooks and docs.
›Renames Knowledge.add_content() and its variants to insert() and insert_many() (old names still work but will be phased out of docs).
└──▷ BREAKING ON UPGRADE
!Removed deprecated fields session_state, dependencies, and user_id from tool functions and hooks where RunContext has replaced them.
!stream_intermediate_steps has been removed; use stream_events instead.
!yield_run_response has been removed; use yield_run_output instead.
!delegate_task_to_all_members has been removed from the Team class.
!tracing_db on AgentOS is deprecated in favor of the new db parameter.
!Instructions are no longer wrapped in <instructions> XML tags by default for Agent and Team; set add_instruction_tags=True to restore the previous behavior.
!DDG web search tool is replaced by WebSearchTools as the default web search tool.
$ git clone --branch 3.1.1 https://github.com/stanfordnlp/dspy.git
# already have the repo? check out this version:$ git checkout 3.1.1
└──▷ USE IT
Persist your current DSPy LM and settings to disk so a pipeline can be restored without reconfiguring.
python
import dspy
dspy.settings.save('dspy_settings.json')
# later, in another process:
dspy.settings.load('dspy_settings.json')
›Adds dspy.RLM module for reinforcement learning-style module execution, backed by an improved PythonInterpreter.
›Adds save and load methods to DSPy settings, enabling persistence and restore of global configuration.
›Adds tool description optimization for multi-agent systems in dspy.gepa, allowing GEPA to tune tool descriptions alongside prompts.
›Enhances StreamListener to support generic type annotations for output, enabling typed streaming results.
›Uses language field in system instructions for dspy.Code fields to guide code-generation formatting.
└──▷ BREAKING ON UPGRADE
!FinalAnswerResult is renamed to FinalOutput in dspy.RLM, and the RLM.__call__ method is removed — code calling RLM as a callable or referencing FinalAnswerResult will break.
Semantic Kernel dotnet-1.69.0 adds FunctionChoiceBehavior support to the Google Gemini connector.
└──▷ GET THIS VERSION
$ git clone --branch dotnet-1.69.0 https://github.com/microsoft/semantic-kernel.git
# already have the repo? check out this version:$ git checkout dotnet-1.69.0
›Adds FunctionChoiceBehavior support to the Google Gemini connector, enabling function/tool calling control for Gemini-backed chat completions.
camel-ai v0.2.83 adds Outlook mail, Lark, IMAP, Serper.dev, AtlasCloud, and Gemma integrations plus new workforce task events
└──▷ GET THIS VERSION
$ git clone --branch v0.2.83 https://github.com/camel-ai/camel.git
# already have the repo? check out this version:$ git checkout v0.2.83
└──▷ USE IT
Write a file and wait for a long-running shell command to finish inside a TerminalToolkit session.
python
from camel.toolkits import TerminalToolkit
toolkit = TerminalToolkit()
# Write a script to disk, then block until a build command completes
toolkit.shell_write_content_to_file(content="echo hello", file_path="/tmp/hello.sh")
toolkit.shell_wait(command="bash /tmp/hello.sh")
Listen for task-lifecycle events — including the new TaskUpdatedEvent — to audit workforce task modifications in real time.
GitHub Copilot CLI v0.0.385 adds infinite sessions via compaction checkpoints, MCP server management with /agent, and combined custom instruction files.
└──▷ GET THIS VERSION
$ git clone --branch v0.0.385 https://github.com/github/copilot-cli.git
# already have the repo? check out this version:$ git checkout v0.0.385
›Enables infinite sessions with automatic long-running context management through compaction checkpoints.
›Adds MCP server management when swapping between custom agents with /agent.
›Combines all custom instruction files instead of using priority-based fallbacks.
›Adds Escape key to cancel a manual /compact command.
›Displays current intent in the terminal tab title.
+1 moreshow less
›Limits store_memory tool inclusion to sessions where memory is enabled for the user.
OpenCode v1.1.26 adds a command execute before hook, raises batch tool limit to 25, and brings vim-style scrolling and mouse permission buttons to the TUI.
└──▷ GET THIS VERSION
$ git clone --branch v1.1.26 https://github.com/sst/opencode.git
# already have the repo? check out this version:$ git checkout v1.1.26
›Adds a command execute before hook, enabling pre-execution scripts to run before any command.
›Increases the maximum batch tool limit from 10 to 25, allowing larger parallel tool calls.
›Adds vim-style line-by-line scrolling keybindings to the TUI.
›Adds mouse support for permission buttons in the TUI.
›Adds an apply_patch tool for OpenAI models.
+10 moreshow less
›Adds a session unshare button to the desktop application.
›Adds an uninstall command for OpenCode installed via Windows package managers.
›Increases question header and label limits in the question tool.
›Auto-routes GPT-5+ models to the Responses API in the GitHub Copilot integration.
›Adds proper variant support to the Copilot provider.
›Treats .fbs files as text instead of images in tool handling.
›Opens the help dialog via the tui/open-help route.
›Centralizes OSC 52 clipboard support for SSH sessions.
›Adds hover overlay with upload and trash icons to project icon in the edit dialog.
›Overhauled Nix flake and packages for improved distribution.
Pi v0.49.2 adds AWS container credential detection, quiet startup, and richer HTML exports with JSONL download.
└──▷ GET THIS VERSION
$ git clone --branch v0.49.2 https://github.com/earendil-works/pi.git
# already have the repo? check out this version:$ git checkout v0.49.2
›Supports AWS credential detection for ECS/Kubernetes environments via AWS_CONTAINER_CREDENTIALS_RELATIVE_URI, AWS_CONTAINER_CREDENTIALS_FULL_URI, and AWS_WEB_IDENTITY_TOKEN_FILE.
›Adds 'quiet startup' setting configurable via /settings.
›HTML exports now include a JSONL download button and jump-to-last-message navigation.
›Adds widgetPlacement option in pi.addWidget() for controlling extension widget placement.
└──▷ BREAKING ON UPGRADE
!The strictResponsesPairing compat option has been removed from the models.json schema.
$ git clone --branch v0.7.2-preview.0 https://github.com/QwenLM/qwen-code.git
# already have the repo? check out this version:$ git checkout v0.7.2-preview.0
›Adds settings support for experimental skills under the experimental namespace in CLI configuration.
›Improves QWEN.md file loading by filtering out system files and limiting scope to relevant directories.
Qwen Code v0.7.1-nightly adds settings support for experimental skills in the CLI.
└──▷ GET THIS VERSION
$ git clone --branch v0.7.1-nightly.20260119.0681c718 https://github.com/QwenLM/qwen-code.git
# already have the repo? check out this version:$ git checkout v0.7.1-nightly.20260119.0681c718
›Adds settings support for experimental skills via the CLI, enabling configuration of opt-in skill features.
THE AI TOOLCHAIN