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.
AutoGPT Platform adds MCP tool block with OAuth, a new flow editor, Telegram blocks, and a POST /graphs external API endpoint.
└──▷ GET THIS VERSION
$ git clone --branch autogpt-platform-beta-v0.6.49 https://github.com/Significant-Gravitas/AutoGPT.git
# already have the repo? check out this version:$ git checkout autogpt-platform-beta-v0.6.49
›Adds POST /graphs endpoint to the external API for programmatic agent graph creation.
›Adds MCP tool block with OAuth support, tool discovery, and standard credential integration.
›Adds new Telegram blocks for Telegram-based automation workflows.
›Replaces the legacy builder with a new flow editor.
›Adds Claude Agent SDK integration for CoPilot interactions.
+13 moreshow less
›Introduces a CoPilot Executor Microservice for running CoPilot tasks.
›Adds workspace file tools, context reconstruction, and transcript upload protection.
›Adds a delete chat session endpoint and corresponding UI.
›Implements a folder organization system for agents.
›Adds SuggestedGoalResponse for handling vague or unachievable goals in CoPilot.
›Adds feature request tools for CoPilot chat.
›Enables parallel tool calls to execute concurrently in CoPilot.
›Adds a stop button wired to cancel executor tasks.
›Adds PDF operations support in the CoPilot executor via fpdf2 dependency.
›Enables WebSearch and consolidates tool constants for CoPilot tooling.
›Adds exact timestamp tooltip on run timestamps in the UI.
›Adds always-visible credentials, inputs, and login prompts in the builder UI.
›Supports workspace:// URLs in regular markdown links.
└──▷ BREAKING ON UPGRADE
!The environment variable LINEAR_API_KEY is renamed to COPILOT_LINEAR_API_KEY; any deployment setting LINEAR_API_KEY for CoPilot will lose access until the variable is renamed.
Haystack v2.25.0 adds SearchableToolset for BM25 tool discovery, a simplified LLM component, and Jinja2-templated Agent prompts.
└──▷ GET THIS VERSION
$ git clone --branch v2.25.0 https://github.com/deepset-ai/haystack.git
# already have the repo? check out this version:$ git checkout v2.25.0
└──▷ USE IT
Let an agent search a large tool catalog at runtime instead of loading every tool into context upfront.
python
from haystack.components.agents import Agent
from haystack.components.generators.chat import OpenAIChatGenerator
from haystack.dataclasses import ChatMessage
from haystack.tools import Tool, SearchableToolset
catalog = [
Tool(name="get_weather", description="Get weather for a city"),
Tool(name="search_web", description="Search the web"),
# ... hundreds more tools
]
toolset = SearchableToolset(catalog=catalog)
agent = Agent(chat_generator=OpenAIChatGenerator(), tools=toolset)
result = agent.run(messages=[ChatMessage.from_user("What's the weather in Milan?")])
Reuse a templated Agent prompt across multiple invocations — useful for translation or summarization pipelines where only the input variable changes.
python
from haystack.components.agents import Agent
from haystack.components.generators.chat import OpenAIChatGenerator
agent = Agent(
chat_generator=OpenAIChatGenerator(),
system_prompt="You are a helpful translation assistant.",
user_prompt="""{% message role="user"%}
Translate the following document to {{ language }}: {{ document }}
{% endmessage %}""",
required_variables=["language", "document"],
)
result = agent.run(language="French", document="The weather is lovely today.")
Use the new LLM component for single-turn, tool-free generation with a templated prompt — ideal for document summarization steps inside a pipeline.
python
from haystack.components.generators.chat import LLM, OpenAIChatGenerator
llm = LLM(
chat_generator=OpenAIChatGenerator(),
system_prompt="You are a helpful assistant.",
user_prompt="""{% message role="user"%}
Summarize the following document: {{ document }}
{% endmessage %}""",
required_variables=["document"],
)
result = llm.run(document="Haystack v2.25.0 introduces SearchableToolset and a new LLM component.")
print(result["last_message"].text)
›Adds SearchableToolset to haystack.tools, enabling agents to dynamically discover tools from large catalogs via BM25 keyword search; starts agents with a single search_tools function and supports configurable search threshold for automatic passthrough mode and top-k result limiting.
›Adds user_prompt and required_variables parameters to the Agent component, enabling reusable Jinja2-templated user prompts that can be passed dynamic variables at runtime without manually constructing ChatMessage objects.
›Adds new LLM component at haystack.components.generators.chat.LLM — a single-turn, tool-free text generation interface supporting system prompts, Jinja2-templated user_prompt, required_variables, streaming callbacks, and both run and run_async execution.
›Adds link_format parameter to PPTXToDocument and XLSXToDocument converters, supporting hyperlink extraction in 'markdown' ([text](url)), 'plain' (text (url)), or 'none' (default, text only) formats.
›Adds FileToFileContent component to convert local files into FileContent objects that can be embedded into ChatMessage for LLM input.
+5 moreshow less
›Adds document_comparison_field parameter to DocumentMRREvaluator, DocumentMAPEvaluator, and DocumentRecallEvaluator, enabling document comparison by fields other than content, including id and metadata keys via meta.<key> syntax.
›Adds support for transformers v5, unlocking faster model loading, improved quantization support, and faster inference for selected models while retaining compatibility with v4.
›Haystack now emits a Warning when dataclass instances (Document, ChatMessage, StreamingChunk, ByteStream, SparseEmbedding) are mutated in place, guiding users toward dataclasses.replace for safe copies.
›LLMDocumentContentExtractor now extracts both content and metadata from image-based documents — when the LLM returns JSON, document_content fills the document body and other keys are merged into metadata; errors are now recorded in extraction_error metadata instead of content_extraction_error.
›EmbeddingBasedDocumentSplitter and MultiQueryEmbeddingRetriever now automatically invoke warm_up() when run() is called if not yet warmed up.
└──▷ BREAKING ON UPGRADE
!The PipelineTemplate and PredefinedPipeline classes and the Pipeline.from_template() method have been removed; migrate to YAML-based pipeline definitions.
!HuggingFaceLocalGenerator default task changed from text2text-generation to text-generation and default model changed from google/flan-t5-base to Qwen/Qwen3-0.6B; existing configs explicitly setting task='text2text-generation' must be updated to task='text-generation' or pin transformers<5.
$ git clone --branch v2026.2.25 https://github.com/openclaw/openclaw.git
# already have the repo? check out this version:$ git checkout v2026.2.25
└──▷ USE IT
Block heartbeat DMs globally while allowing specific agents to still deliver direct messages.
yaml
# In your OpenClaw config file:
agents:
defaults:
heartbeat:
directPolicy: "block"
list:
- id: my-agent
heartbeat:
directPolicy: "allow"
›New agents.defaults.heartbeat.directPolicy config key (allow | block) replaces the old heartbeat DM toggle, with per-agent override via agents.list[].heartbeat.directPolicy.
›Adds startup macrobenchmark and low-noise perf CLI scripts for deterministic cold-start tracking on Android.
›Adds mobile stacked layout for compose action buttons on small screens in the chat UI.
└──▷ BREAKING ON UPGRADE
!The heartbeat DM toggle is replaced by agents.defaults.heartbeat.directPolicy; the default is now allow. To preserve the DM-blocked behavior from v2026.2.24, set agents.defaults.heartbeat.directPolicy: "block" (or use a per-agent override via agents.list[].heartbeat.directPolicy).
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 v2.1.59 adds auto-memory saving, a /copy picker for code blocks, and smarter always-allow bash prefixes.
└──▷ GET THIS VERSION
$ git clone --branch v2.1.59 https://github.com/anthropics/claude-code.git
# already have the repo? check out this version:$ git checkout v2.1.59
└──▷ TRY IT
After a session, review and edit what Claude auto-saved to persistent memory so future sessions inherit the right context.
$ /memory
When a response contains multiple code blocks, pick exactly the one snippet you want on your clipboard instead of manually selecting text.
$ /copy
›Adds auto-memory: Claude automatically saves useful context to persistent memory, manageable via the /memory command.
›New /copy command opens an interactive picker to select and copy individual code blocks or the full response.
›Smarter 'always allow' prefix suggestions for compound bash commands — computes per-subcommand prefixes (e.g. each part of cd /tmp && git fetch && git push) instead of treating the whole command as one unit.
›Reduces memory usage in multi-agent sessions by releasing completed subagent task state.
Crush v0.46.0 adds MiniMax China provider and granular Anthropic reasoning levels for Opus/Sonnet 4.6.
└──▷ GET THIS VERSION
$ git clone --branch v0.46.0 https://github.com/charmbracelet/crush.git
# already have the repo? check out this version:$ git checkout v0.46.0
›Adds MiniMax China as a new provider, using the China-specific endpoint for Chinese users.
›Adds granular reasoning/thinking effort levels for Anthropic Opus 4.6 and Sonnet 4.6, replacing the previous on/off toggle with adaptive level selection.
›Adds local inference provider backed by llama.cpp with HuggingFace model management.
›Adds Cerebras provider support.
›Adds Moonshot and Kimi Code as declarative providers.
›Adds LMStudio as a declarative provider.
›Adds a gateway to chat with Goose via Telegram and similar messaging platforms.
+17 moreshow less
›Adds MCP apps sampling support.
›Adds Neighborhood extension to the Extensions Library.
›Overhauled computer controller with Peekaboo integration.
›Adds a TUI client for goose-acp.
›Adds GOOSE_SUBAGENT_MODEL and GOOSE_SUBAGENT_PROVIDER config options to control sub-agent model selection.
›Adds configurable OpenTelemetry logging level.
›Adds GoosePlatform in AgentConfig and MCP initialization.
›Adds Gemini CLI streaming support via stream-json events.
›Exposes context window utilization to the agent via MOIM.
›Adds auto-submit for recipes that have been accepted.
›Adds Claude Code permission prompt routing for approve mode.
›Displays token counts directly for 'free' providers.
›Adds a TypeScript SDK for ACP extension methods.
›Adds Bedrock prompt cache support.
›Removes allows_unlisted_models flag — custom model entry is now always permitted.
›Displays working directory in the UI.
›Adds local model settings access from the bottom-bar model menu.
└──▷ BREAKING ON UPGRADE
!The allows_unlisted_models flag has been removed; custom model entry is now unconditionally enabled and any configuration relying on this flag will no longer be respected.
›Adds a Kotlin function for getting structured outputs.
└──▷ BREAKING ON UPGRADE
!Tool router environment variable names have been updated to include a goose prefix (e.g., existing env vars without the prefix will no longer be recognized).
Lightweight coding agent that runs in your terminal
Codex CLI gains a direct install script, js_repl promotion, user-input in Default mode, and smarter memory management.
└──▷ GET THIS VERSION
$ git clone --branch rust-v0.106.0 https://github.com/openai/codex.git
# already have the repo? check out this version:$ git checkout rust-v0.106.0
›Adds a direct install script for macOS and Linux, published as a GitHub release asset alongside the codex and rg binaries.
›Promotes js_repl to /experimental with startup compatibility checks, user-visible warnings for incompatible Node versions, and a lowered minimum Node requirement of 22.22.0.
›Enables request_user_input in Default collaboration mode, not only in Plan mode.
›Makes the 5.3-codex model visible in the CLI model list for API users.
›Introduces diff-based memory forgetting and usage-aware memory selection for improved long-session memory behavior.
+3 moreshow less
›Expands the app-server v2 thread API with experimental thread-scoped realtime endpoints/notifications and a thread/unsubscribe flow to unload live threads without archiving them.
›Adds structured OTEL audit logging for embedded codex-network-proxy policy decisions and blocks.
›Reduces sub-agent startup overhead by skipping expensive history metadata scans during subagent spawns.
Phoenix 13.4.0 adds a refusal evaluator, a GraphQL CLI command, PHOENIX_MIGRATE_INDEX_CONCURRENTLY flag, and SQL stdout visibility.
└──▷ GET THIS VERSION
$ git clone --branch arize-phoenix-v13.4.0 https://github.com/Arize-ai/phoenix.git
# already have the repo? check out this version:$ git checkout arize-phoenix-v13.4.0
└──▷ TRY IT
Run a rolling DB migration without locking tables — set before starting Phoenix in a multi-replica deployment.
$ export PHOENIX_MIGRATE_INDEX_CONCURRENTLY=true
Query the Phoenix GraphQL API from the CLI without opening a browser or constructing a raw curl request.
$ px api graphql
›Adds PHOENIX_MIGRATE_INDEX_CONCURRENTLY environment variable to enable concurrent index migrations for rolling deployments without downtime.
›Adds px api graphql subcommand to the Phoenix CLI for issuing GraphQL queries directly from the command line.
›Adds ability to print SQL to stdout for observability into database query activity.
›Adds a refusal evaluator to the evals module for detecting LLM refusal responses.
›Adds navigation counters in the UI for projects, datasets, evaluators, and prompts to surface item counts at a glance.
Composio CLI 0.1.28 enhances init and tool-router based tool discovery with improved tool search and API key inference.
└──▷ GET THIS VERSION
$ git clone --branch @composio/[email protected] https://github.com/ComposioHQ/composio.git
# already have the repo? check out this version:$ git checkout @composio/[email protected]
›Enhances init and tool-router based tool discovery in the CLI for improved tool search and API key inference.
THE AI TOOLCHAIN