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 v2.5.16 adds LLMsTxtTools, SalesforceTools, Azure AI Foundry Claude, and OpenAI Responses background mode
└──▷ GET THIS VERSION
$ git clone --branch v2.5.16 https://github.com/agno-agi/agno.git
# already have the repo? check out this version:$ git checkout v2.5.16
›Adds LLMsTxtTools and LLMsTxtReader classes for consuming the llms.txt standard, enabling agents to index LLM-friendly documentation from sites that expose a /llms.txt endpoint (e.g. https://docs.agno.com/llms.txt).
›Adds SalesforceTools for integrating Salesforce CRM data and actions into agents.
›Adds Azure AI Foundry Claude as a new model provider.
›Adds background mode support for the OpenAI Responses API.
LangChain Core 1.3.0a1 adds ContextOverflowError, multimodal token counting, XML buffer format, tool-call metadata, and more new APIs.
└──▷ GET THIS VERSION
$ git clone --branch langchain-core==1.3.0a1 https://github.com/langchain-ai/langchain.git
# already have the repo? check out this version:$ git checkout langchain-core==1.3.0a1
└──▷ USE IT
Catch context-window overflows explicitly instead of catching a generic exception, so you can retry with a shorter prompt.
python
from langchain_core.exceptions import ContextOverflowError
try:
response = llm.invoke(messages)
except ContextOverflowError:
messages = messages[-10:] # trim history and retry
response = llm.invoke(messages)
Serialize a chat history to XML for downstream XML-aware processing or storage.
python
from langchain_core.messages import get_buffer_string
xml_history = get_buffer_string(messages, format='xml')
print(xml_history)
›Adds ContextOverflowError exception class (raised automatically in Anthropic and OpenAI integrations when context window is exceeded).
›Adds 'approximate' as an alias for count_tokens_approximately in token-counting calls.
›Adds count_tokens_approximately support for tool schemas — token estimates now include tool definitions.
›Adds multimodal support to count_tokens_approximately — image and other non-text message content is now included in approximate token counts.
›Adds scaling by reported usage in count_tokens_approximately to improve accuracy against real model outputs.
+15 moreshow less
›Adds usage_metadata field to metadata emitted by LangChainTracer, making token-usage data visible in LangSmith traces.
›Adds tool_call_count automatic counting and storage in message metadata.
›Adds tool_call_id to on_tool_error event data for better error attribution in callbacks.
›Adds XML format option to get_buffer_string() for serializing conversation history as XML.
›Adds custom message separator support to get_buffer_string() via a new separator argument.
›Adds text_inputs and text_outputs fields to model-profiles.
›Adds PEP 702 __deprecated__ attribute support to the @deprecated decorator.
›Adds LangSmith integration metadata to create_agent and init_chat_model.
›Adds ChatBaseten to the serializable mapping for persistence and tracing.
›Adds anti-SSRF hardening to langchain-core.
›Adds more file extensions to the ignore list in HTML link extraction utilities.
›Adds langchain-openrouter as a new provider package.
›Adds BaseCrossEncoder to langchain-core.
›Adds base_url configuration support documented in the Mermaid API diagramming integration.
›Adds imputed placeholder filenames for OpenAI file inputs when no filename is supplied.
PydanticAI v1.80.0 adds capability ordering, hooks ordering, and server-side context compaction for OpenAI and Anthropic
└──▷ GET THIS VERSION
$ git clone --branch v1.80.0 https://github.com/pydantic/pydantic-ai.git
# already have the repo? check out this version:$ git checkout v1.80.0
└──▷ USE IT
Use OpenAICompaction to automatically compact context on the server side when approaching token limits with an OpenAI model.
python
from pydantic_ai.capabilities import OpenAICompaction
agent = Agent(
'openai:gpt-4o',
capabilities=[OpenAICompaction()],
)
Declare that one capability must wrap another using CapabilityOrdering to enforce a guaranteed composition order.
python
from pydantic_ai.capabilities import CapabilityOrdering
ordering = CapabilityOrdering(my_outer_capability, wraps=my_inner_capability)
›Adds CapabilityOrdering with relationship descriptors innermost, outermost, wraps, wrapped_by, and requires to control how capabilities compose and resolve ordering.
›Adds an ordering parameter to Hooks and supports instance references in wraps/wrapped_by for finer control over hook execution order.
›Adds OpenAICompaction and AnthropicCompaction capability classes to enable server-side context window compaction for those providers.
PydanticAI v1.79.0 adds AG-UI 0.1.13/0.1.15 support, a new async HTTP client factory, and apply() on capability classes.
└──▷ GET THIS VERSION
$ git clone --branch v1.79.0 https://github.com/pydantic/pydantic-ai.git
# already have the repo? check out this version:$ git checkout v1.79.0
└──▷ USE IT
Use create_async_http_client as a context manager to control the lifetime of the shared async HTTP client explicitly.
python
from pydantic_ai.http import create_async_http_client
async with create_async_http_client() as client:
agent = MyAgent(http_client=client)
result = await agent.run('Hello')
›Adds create_async_http_client context manager to replace the internal HTTP client cache, giving callers explicit lifecycle control over async HTTP clients.
›Adds apply() method to AbstractCapability, CombinedCapability, and WrapperCapability, enabling capabilities to be applied directly.
›Adds full AG-UI 0.1.13 and 0.1.15 support, including reasoning, multi-modal messaging, and dump_messages.
Claude Code is an agentic coding tool that lives in your terminal, understands your codebase, and helps you code faster by executing routine tasks, explaining complex code, and handling git workflows - all through natural language commands.
Claude Code v2.1.101 adds /team-onboarding, OS CA store trust by default, and auto-create cloud environments for /ultraplan.
└──▷ GET THIS VERSION
$ git clone --branch v2.1.101 https://github.com/anthropics/claude-code.git
# already have the repo? check out this version:$ git checkout v2.1.101
└──▷ TRY IT
Generate a ramp-up guide for a new teammate joining your project, based on your existing Claude Code usage patterns.
$ /team-onboarding
Resume a named session by its human-readable title (set via /rename) rather than an opaque session ID.
$ claude -p--resume my-project-audit
Restrict TLS trust to bundled CAs only, opting out of the new OS CA store default (e.g., in locked-down environments where OS trust anchors should not be used).
$ CLAUDE_CODE_CERT_STORE=bundled claude
›Adds /team-onboarding command to generate a teammate ramp-up guide from your local Claude Code usage.
›Trusts OS CA certificate store by default, enabling enterprise TLS proxies without extra configuration (set CLAUDE_CODE_CERT_STORE=bundled to restrict to bundled CAs only).
›Enables /ultraplan and other remote-session features to auto-create a default cloud environment, removing the requirement to complete web setup first.
›Beta tracing now honors OTEL_LOG_USER_PROMPTS, OTEL_LOG_TOOL_DETAILS, and OTEL_LOG_TOOL_CONTENT; sensitive span attributes are opt-in only.
›claude -p --resume <name> now accepts session titles set via /rename or --name.
+5 moreshow less
›Rate-limit retry messages now show which limit was hit and when it resets.
›Refusal error messages now include the API-provided explanation when available.
›Tool-not-available errors now explain why the tool is unavailable and how to proceed.
›/plugin and claude plugin update now warn when the marketplace could not be refreshed instead of silently showing stale versions.
›SDK query() now cleans up subprocesses and temp files when consumers break from for await or use await using.
└──▷ BREAKING ON UPGRADE
!OS CA certificate store is now trusted by default; set CLAUDE_CODE_CERT_STORE=bundled to revert to bundled CAs only.
Copilot CLI v1.0.24: hooks now propagate modifiedArgs and custom agents accept VS Code display names
└──▷ GET THIS VERSION
$ git clone --branch v1.0.24 https://github.com/github/copilot-cli.git
# already have the repo? check out this version:$ git checkout v1.0.24
›preToolUse hooks now correctly propagate modifiedArgs/updatedInput and additionalContext fields, enabling hooks to pass modified arguments and extra context downstream.
›Custom agent model field now accepts VS Code display names and vendor suffixes (e.g., Claude Sonnet 4.5, GPT-5.4 (copilot)) in addition to raw model IDs.
›The --remote flag is now respected when the session sync prompt appears on first run inside a GitHub repo.
›Redesigned exit screen with Copilot mascot and a cleaner usage summary layout.
Lightweight coding agent that runs in your terminal
Codex CLI v0.119.0 adds voice v2 WebRTC, codex exec-server, Ctrl+O copy, /resume by ID, and richer MCP server support.
└──▷ GET THIS VERSION
$ git clone --branch rust-v0.119.0 https://github.com/openai/codex.git
# already have the repo? check out this version:$ git checkout rust-v0.119.0
›Realtime voice sessions now default to the v2 WebRTC path with configurable transport, voice selection, and native TUI media support.
›New experimental codex exec-server subcommand for remote/app-server workflows.
›Remote app-server sessions now support egress WebSocket transport and remote --cd forwarding.
›Press Ctrl+O in the TUI to copy the latest agent response to the clipboard, with improved SSH and cross-platform clipboard behavior.
›/resume can now jump directly to a session by ID or name from the TUI session picker.
+5 moreshow less
›TUI notifications are more configurable, adding Warp OSC 9 support and an opt-in mode for notifications even while the terminal is focused.
›MCP Apps and custom MCP servers gain support for resource reads, tool-call metadata, custom-server tool search, server-driven elicitations, and file-parameter uploads.
›Runtime remote-control enablement and sandbox-aware filesystem APIs added for app-server workflows.
›Enables disabling prompt instruction blocks and environment context injection via config.
›Fuzzy file search is now case-insensitive in the TUI.
└──▷ BREAKING ON UPGRADE
!The OPENAI_BASE_URL config fallback has been removed.
Phoenix v14.2.0 adds name-based project URL routing via /redirects/projects/:project_name.
└──▷ GET THIS VERSION
$ git clone --branch arize-phoenix-v14.2.0 https://github.com/Arize-ai/phoenix.git
# already have the repo? check out this version:$ git checkout arize-phoenix-v14.2.0
└──▷ TRY IT
Navigate directly to a project by name instead of looking up its internal ID — useful for bookmarks, dashboards, or sharing links with teammates.
›Adds /redirects/projects/:project_name endpoint for resolving projects by name rather than by internal ID, enabling stable, human-readable project URLs.
THE AI TOOLCHAIN