The AI Toolchain — issue -364, July 31, 2025

Agno v1.7.6 adds Portkey models, BitbucketTools, JinaEmbedder, EvmTools, LinkupTools, RowChunking, and non-blocking Workflows 2.0 background execution.

$ git clone --branch v1.7.6 https://github.com/agno-agi/agno.git
# already have the repo? check out this version:
$ git checkout v1.7.6

from agno.document.chunking.row import RowChunking
from agno.document.reader.csv_reader import CSVReader

reader = CSVReader(chunking_strategy=RowChunking())
documents = reader.read('alerts.csv')

from agno.tools.bitbucket import BitbucketTools
from agno.agent import Agent

agent = Agent(
    tools=[BitbucketTools(username="<username>", password="<app-password>", workspace="<workspace>")],
    markdown=True,
)
agent.print_response("List all open pull requests in the agno repo")

• ›Adds BitbucketTools class for interacting with Bitbucket Cloud repository APIs from an agent.
• ›Adds JinaEmbedder class for using Jina-hosted embedding models.
• ›Adds EvmTools class for executing transactions on EVM-compatible blockchains via the web3 library.
• ›Adds LinkupTools class for web search capabilities inside agents.
• ›Adds RowChunking as a CSV-specific chunking strategy for document ingestion.

Agno v1.7.5 adds SurrealDB as a vector DB backend and cache_session control for memory management.

$ git clone --branch v1.7.5 https://github.com/agno-agi/agno.git
# already have the repo? check out this version:
$ git checkout v1.7.5

• ›Adds cache_session attribute to agent/session configuration, allowing users to disable session caching for improved memory management.
• ›Adds SurrealDB support as a vector database backend for knowledge bases.
• ›Adds Workflows 2.0 support inside FastAPIApp, enabling the new workflow engine to run as a FastAPI application.

Agno v1.7.4 ships a redesigned step-based Workflows 2.0 (beta) and Pydantic model input support for Agent and Team.

$ git clone --branch v1.7.4 https://github.com/agno-agi/agno.git
# already have the repo? check out this version:
$ git checkout v1.7.4

from pydantic import BaseModel
from agno.agent import Agent

class ScanRequest(BaseModel):
    target: str
    depth: int

agent = Agent(model=...)
agent.run(ScanRequest(target="example.com", depth=3))

• ›Adds Workflows 2.0 (beta), a complete redesign of the workflow system using a step-based architecture that supports sequential, parallel, conditional, and loop-based execution, dynamic step routing, mixed components (agents, teams, and functions), and shared session state across steps.
• ›Both Agent and Team now accept a Pydantic model as structured input on run() and print_response().

Agno v1.7.3 adds session_state on agent/team runs and GCSPDFKnowledgeBase for Google Cloud Storage PDFs.

$ git clone --branch v1.7.3 https://github.com/agno-agi/agno.git
# already have the repo? check out this version:
$ git checkout v1.7.3

agent.run('Continue the investigation', session_state={'case_id': 'INC-4821', 'severity': 'high'})

• ›Adds GCSPDFKnowledgeBase class to load and query PDFs stored on Google Cloud Storage as a knowledge base source.
• ›Adds session_state parameter to agent and team run calls, allowing callers to pass initial session state at invocation time.

Agno v1.7.2 adds MySQLStorage backend, XAi live search, OpenAI deep research models, and memory growth tracking.

$ git clone --branch v1.7.2 https://github.com/agno-agi/agno.git
# already have the repo? check out this version:
$ git checkout v1.7.2

from agno.eval.performance import PerformanceEval
from agno.agent import Agent

eval = PerformanceEval(agent=Agent(), memory_growth_tracking=True)
eval.run()

from agno.agent import Agent
from agno.models.openai import OpenAIChat

agent = Agent(model=OpenAIChat(id="o3-deep-research"))
agent.print_response("Research the latest advances in quantum error correction.")

• ›Adds MySQLStorage class as a session storage backend for agents, teams, and workflows.
• ›Adds memory_growth_tracking attribute on PerformanceEval to enable debug logs for memory growth during performance evaluations.
• ›Adds agent and team as optional parameters in tool hooks for greater flexibility.
• ›Supports live search on the XAi model provider.
• ›Supports o4-mini-deep-research and o3-deep-research OpenAI model identifiers.

Agno v1.7.1 adds debug_level to Agent/Team, Gemini thinking params, Valyu/Oxylabs toolkits, and new Serper tools.

$ git clone --branch v1.7.1 https://github.com/agno-agi/agno.git
# already have the repo? check out this version:
$ git checkout v1.7.1

from agno.agent import Agent

agent = Agent(
    model=my_model,
    debug_level=2,
)

from agno.models.gemini import Gemini

model = Gemini(
    id="gemini-2.0-flash-thinking-exp",
    thinking_budget=1024,
    include_thoughts=True,
)

from agno.agent import Agent
from agno.tools.valyu import ValyuTools

agent = Agent(tools=[ValyuTools()])
agent.print_response("Find recent papers on retrieval-augmented generation")

• ›Adds debug_level parameter (int 1 or 2) to both Agent and Team classes for controlling logging verbosity, with 2 enabling more verbose model logs.
• ›Adds thinking_budget and include_thoughts parameters to the Gemini model class for configuring Gemini thinking behavior.
• ›Adds parser_model parameter support to Team for structured output via a dedicated parser model.
• ›Adds search_news, search_scholar, and scrape_webpage tools to the Serper toolkit.
• ›New OxylabsTools toolkit for web-scraping capabilities in agents.

AutoGPT Platform gains Excel support, Gmail thread blocks, Replicate model blocks, GCS file storage, and a new ReverseListOrderBlock.

$ git clone --branch autogpt-platform-beta-v0.6.17 https://github.com/Significant-Gravitas/AutoGPT.git
# already have the repo? check out this version:
$ git checkout autogpt-platform-beta-v0.6.17

• ›Adds Excel file support to ReadSpreadsheetBlock and introduces a new FileReadBlock for reading files within agent workflows.
• ›Adds Gmail thread blocks for reading and interacting with Gmail threads in agent pipelines.
• ›Adds Replicate model blocks, enabling agents to call Replicate-hosted AI models.
• ›Integrates GCS (Google Cloud Storage) file storage with automatic expiration for Agent File Input.
• ›Adds ReverseListOrderBlock for reversing the order of list elements in agent pipelines.

AutoGPT Platform adds a Block Development SDK with auto-registration, block error rate monitoring with Discord alerts, and builder credentials support.

$ git clone --branch autogpt-platform-beta-v0.6.16 https://github.com/Significant-Gravitas/AutoGPT.git
# already have the repo? check out this version:
$ git checkout autogpt-platform-beta-v0.6.16

• ›Adds a Block Development SDK with an auto-registration system for building and registering custom blocks.
• ›Adds block error rate monitoring with Discord alerts to surface runtime block failures.
• ›Adds builder credentials support with UX improvements for managing credentials in the builder.
• ›New navbar design and updated UI for logged-out pages.
• ›Optimizes StoreAgent and Creator views with database indexes and materialized views for improved query performance.

AutoGPT Platform v0.6.15 adds KV storage blocks, context-aware prompt compaction, Perplexity Sonar models, and expanded data/media blocks.

$ git clone --branch autogpt-platform-beta-v0.6.15 https://github.com/Significant-Gravitas/AutoGPT.git
# already have the repo? check out this version:
$ git checkout autogpt-platform-beta-v0.6.15

• ›Adds aiohttp.BasicAuth support in make_request for blocks making HTTP requests with Basic Auth credentials.
• ›Introduces GetPersonDetailBlock and enriches SearchPeopleBlock with email enrichment for people-lookup workflows.
• ›Implements KV data storage blocks, enabling key-value read/write operations within agent graphs.
• ›Introduces context-window-aware prompt compaction for LLM and SmartDecision blocks to handle large inputs without manual truncation.
• ›Improves CreateListBlock to support batching based on token count, preventing LLM context overflows in list-generation workflows.

CrewAI 0.150.0 adds ad-hoc tool calling, Mem0 v2 storage, SerperScrapeWebsiteTool, and Bedrock AgentCore browser/code interpreter toolkits.

$ git clone --branch 0.150.0 https://github.com/crewAIInc/crewAI.git
# already have the repo? check out this version:
$ git checkout 0.150.0

• ›Adds support for ad-hoc tool calling via the internal LLM class, enabling on-demand tool invocation outside of standard crew/agent flows.
• ›Upgrades Mem0 Storage integration from v1.1 to v2.
• ›New SerperScrapeWebsiteTool extracts clean content from URLs using Serper.
• ›Integrates Bedrock AgentCore browser and code interpreter toolkits for use with Bedrock agents.
• ›Adds UserMemory deprecation notice to signal future removal.

CrewAI 0.148.0 introduces Agent evaluation functionality with thread-safe AgentEvaluator and neatlogs integration.

$ git clone --branch 0.148.0 https://github.com/crewAIInc/crewAI.git
# already have the repo? check out this version:
$ git checkout 0.148.0

• ›Introduces Agent evaluation functionality with AgentEvaluator, supporting regression testing and experiment methods for both Agent and LiteAgent.
• ›Enables event emission during Agent evaluation for observability into evaluation runs.
• ›Adds crew context tracking for LLM guardrail events.
• ›Adds integration with neatlogs for structured agent log management.

CrewAI 0.141.0 adds crew context tracking for LLM guardrail events.

$ git clone --branch 0.141.0 https://github.com/crewAIInc/crewAI.git
# already have the repo? check out this version:
$ git checkout 0.141.0

• ›Adds crew context tracking for LLM guardrail events, enabling richer auditability of guardrail decisions within a crew's execution context.

CrewAI 0.140.0 adds LLM call tracking by task/agent, MemoryEvents monitoring, and a WorkOS CLI login command.

$ git clone --branch 0.140.0 https://github.com/crewAIInc/crewAI.git
# already have the repo? check out this version:
$ git checkout 0.140.0

• ›Tracks LLM calls broken down by task and agent for observability into crew execution costs and patterns.
• ›Introduces MemoryEvents to monitor memory usage within a crew run.
• ›Adds console logging for memory system and LLM guardrail events.
• ›Improves data training support for models up to 7B parameters.
• ›Adds workos login command to the CLI for WorkOS-based authentication.

DSPy 3.0.0b2 adds reusable stream listeners, PEP 604 union types in signatures, Gemini provider support, and format control for ToolCalls.

$ git clone --branch 3.0.0b2 https://github.com/stanfordnlp/dspy.git
# already have the repo? check out this version:
$ git checkout 3.0.0b2

import dspy

class Classify(dspy.Signature):
    text: str = dspy.InputField()
    label: int | str = dspy.OutputField()

predictor = dspy.Predict(Classify)
result = predictor(text='The sky is blue')

import dspy

lm = dspy.LM('gemini/gemini-1.5-pro')
dspy.configure(lm=lm)

• ›Adds format parameter to ToolCalls for controlling tool call output format.
• ›Supports PEP 604 union types (e.g. int | str) in inline signatures, enabling modern Python type hint syntax in dspy.Signature definitions.
• ›Adds Gemini as a supported LM provider.
• ›Changes default model for the Databricks provider to llama-4.
• ›Allows reusing the StreamListener across multiple streaming calls.

LangChain text-splitters 0.3.9 adds Visual Basic 6 language support and a keep_separator option for HTMLSemanticPreservingSplitter.

$ git clone --branch langchain-text-splitters==0.3.9 https://github.com/langchain-ai/langchain.git
# already have the repo? check out this version:
$ git checkout langchain-text-splitters==0.3.9

from langchain_text_splitters import HTMLSemanticPreservingSplitter

splitter = HTMLSemanticPreservingSplitter(keep_separator=True)
chunks = splitter.split_text(html_content)

• ›Adds keep_separator argument to HTMLSemanticPreservingSplitter, letting callers control whether HTML separators are retained in output chunks.
• ›Adds chunk_size and chunk_overlap validation, raising errors early when invalid splitter parameters are supplied.
• ›Adds Visual Basic 6 as a supported language for code-aware text splitting.
• ›Hardens XML parsing in HTMLSectionSplitter by removing the xslt_path parameter and tightening the parser configuration.

langchain-perplexity 0.1.2 exposes search_results from the Perplexity chat model response.

$ git clone --branch langchain-perplexity==0.1.2 https://github.com/langchain-ai/langchain.git
# already have the repo? check out this version:
$ git checkout langchain-perplexity==0.1.2

• ›Exposes search_results field in the Perplexity chat model response, giving callers direct access to the web sources Perplexity used to ground its answer.

LangChain Core 0.3.71 adds a sanitize_for_postgres utility to prevent PostgreSQL NUL byte errors.

$ git clone --branch langchain-core==0.3.71 https://github.com/langchain-ai/langchain.git
# already have the repo? check out this version:
$ git checkout langchain-core==0.3.71

• ›Adds sanitize_for_postgres utility function to strip NUL bytes from data before PostgreSQL writes, preventing DataError exceptions.

langchain-chroma 0.2.5 adds Chroma Cloud support to the LangChain vector store integration.

$ git clone --branch langchain-chroma==0.2.5 https://github.com/langchain-ai/langchain.git
# already have the repo? check out this version:
$ git checkout langchain-chroma==0.2.5

• ›Adds Chroma Cloud support, enabling the Chroma vector store to connect to Chroma's managed cloud offering.

langchain-ollama 0.3.6 warns on empty load responses for faster debugging.

$ git clone --branch langchain-ollama==0.3.6 https://github.com/langchain-ai/langchain.git
# already have the repo? check out this version:
$ git checkout langchain-ollama==0.3.6

• ›Adds a warning when Ollama returns empty load responses, surfacing silent model-loading failures at runtime.

langchain-huggingface 0.3.1 adds support for the image-text-to-text pipeline task.

$ git clone --branch langchain-huggingface==0.3.1 https://github.com/langchain-ai/langchain.git
# already have the repo? check out this version:
$ git checkout langchain-huggingface==0.3.1

• ›Adds support for the image-text-to-text pipeline task in HuggingFace pipelines.

LangChain Core 0.3.69 adds permissive deserialization mode and integer merging when combining dicts.

$ git clone --branch langchain-core==0.3.69 https://github.com/langchain-ai/langchain.git
# already have the repo? check out this version:
$ git checkout langchain-core==0.3.69

• ›Adds an option to make deserialization more permissive, allowing looser loading of serialized objects.
• ›Supports integer value combining when merging dicts, enabling numeric fields to be summed rather than overwritten during merge operations.

ChatGroq gains a service tier option for controlling request priority or cost in langchain-groq 0.3.6.

$ git clone --branch langchain-groq==0.3.6 https://github.com/langchain-ai/langchain.git
# already have the repo? check out this version:
$ git checkout langchain-groq==0.3.6

from langchain_groq import ChatGroq

llm = ChatGroq(
    model="llama3-70b-8192",
    service_tier="flex"
)

• ›Adds service_tier option to ChatGroq to control the service tier used for Groq API requests.

langchain-ollama 0.3.4 adds thinking/reasoning mode, tool-call streaming, and model validation on init.

$ git clone --branch langchain-ollama==0.3.4 https://github.com/langchain-ai/langchain.git
# already have the repo? check out this version:
$ git checkout langchain-ollama==0.3.4

from langchain_ollama import ChatOllama

llm = ChatOllama(model="llama3", validate_model_on_init=True)

• ›Adds validate_model_on_init option to catch model configuration errors at initialization time rather than at inference.
• ›Supports Ollama thinking/reasoning mode, configurable per-call so individual invocations can enable or disable reasoning independently.
• ›Enables tool-call streaming for Ollama-backed chains and agents.

langchain-mistralai now includes finish_reason in response metadata when parsing streaming chunks.

$ git clone --branch langchain-mistralai==0.2.11 https://github.com/langchain-ai/langchain.git
# already have the repo? check out this version:
$ git checkout langchain-mistralai==0.2.11

• ›Adds finish_reason to response metadata when parsing MistralAI chunks into AIMessageChunk, making stop-reason inspection available on streamed responses.

langchain-groq 0.3.5 adds reasoning_effort parameter support for ChatGroq models.

$ git clone --branch langchain-groq==0.3.5 https://github.com/langchain-ai/langchain.git
# already have the repo? check out this version:
$ git checkout langchain-groq==0.3.5

from langchain_groq import ChatGroq

llm = ChatGroq(model="deepseek-r1-distill-llama-70b", reasoning_effort="default")
response = llm.invoke("Explain the halting problem.")
print(response.content)

• ›Adds reasoning_effort parameter to ChatGroq for controlling model reasoning depth on supported Groq models.

LangGraph prebuilt 0.6.0 adds dynamic model selection in create_react_agent and a new context API replacing config['configurable'].

$ git clone --branch prebuilt==0.6.0 https://github.com/langchain-ai/langgraph.git
# already have the repo? check out this version:
$ git checkout prebuilt==0.6.0

• ›Adds dynamic model support to create_react_agent, allowing the LLM to be swapped at runtime per invocation.
• ›Introduces a new context API as a cleaner replacement for config['configurable'] and config_schema patterns.

LangGraph CLI 0.3.6 introduces an api-version option and a new context API replacing config['configurable'] and config_schema.

$ git clone --branch cli==0.3.6 https://github.com/langchain-ai/langgraph.git
# already have the repo? check out this version:
$ git checkout cli==0.3.6

• ›Adds api-version option for explicit API version control.
• ›Introduces new context API as a replacement for config['configurable'] and config_schema for passing configuration to graph nodes.

LangGraph Python SDK 0.2.0 adds context API support and exposes interrupts in thread state

$ git clone --branch sdk==0.2.0 https://github.com/langchain-ai/langgraph.git
# already have the repo? check out this version:
$ git checkout sdk==0.2.0

thread_state = await client.threads.get_state(thread_id)
interrupts = thread_state.interrupts

• ›Adds SDK support for the context API, enabling callers to pass context through the LangGraph SDK.
• ›Adds interrupts field to thread state, making interrupt information accessible when inspecting thread state.
• ›Cleans up the Interrupt interface for v1, refining the interrupt contract.

LangGraph 0.5.4 adds ParentCommand handling in RemoteGraph for cross-graph command propagation.

$ git clone --branch 0.5.4 https://github.com/langchain-ai/langgraph.git
# already have the repo? check out this version:
$ git checkout 0.5.4

• ›Supports ParentCommand in RemoteGraph, enabling commands issued inside a remote graph to propagate up to the parent graph.

LangGraph SDK 0.1.73 exposes is_studio_user flag to identify Studio-originated requests.

$ git clone --branch sdk==0.1.73 https://github.com/langchain-ai/langgraph.git
# already have the repo? check out this version:
$ git checkout sdk==0.1.73

• ›Adds is_studio_user attribute to identify whether the current user is a LangGraph Studio user.

LangGraph checkpoint-postgres 2.0.22 adds numpy array serialization and pandas pickle fallback in JsonPlusSerializer.

$ git clone --branch checkpointpostgres==2.0.22 https://github.com/langchain-ai/langgraph.git
# already have the repo? check out this version:
$ git checkout checkpointpostgres==2.0.22

• ›Supports numpy array serialization in JsonPlusSerializer, enabling checkpoint storage of numpy arrays without manual conversion.
• ›Adds pickle fallback for pandas objects in JsonPlusSerializer via serialize/deserialize path, so DataFrames and Series round-trip through checkpoints reliably.
• ›Extends pipeline mode in checkpoint-postgres to use the same lock used in non-pipeline mode, improving consistency under concurrent writes.
• ›Centralizes CheckpointTuple creation into a shared helper function within checkpoint_postgres, reducing duplication across sync and async paths.

LangGraph CLI 0.3.4 adds a flag to retain build dependencies (setuptools, pip, wheel) in container builds.

$ git clone --branch cli==0.3.4 https://github.com/langchain-ai/langgraph.git
# already have the repo? check out this version:
$ git checkout cli==0.3.4

• ›Adds a CLI argument to retain build dependencies (setuptools, pip, wheel) in the build output instead of pruning them.

Letta 0.8.9 adds multi-provider summarization, agent loop cancellation, and MCP custom headers

$ git clone --branch 0.8.9 https://github.com/letta-ai/letta.git
# already have the repo? check out this version:
$ git checkout 0.8.9

• ›Supports custom headers for MCP (Model Context Protocol) connections.
• ›Enables agent loop run cancellation, allowing in-flight agent executions to be stopped.
• ›Supports configuring different providers for summarization, decoupling summary generation from the primary model provider.
• ›Improvements to file management capabilities.

LlamaIndex v0.12.52 adds a Jira issue tool spec, web reader timeouts, and optimized BGEM3Index persistence.

$ git clone --branch v0.12.52 https://github.com/run-llama/llama_index.git
# already have the repo? check out this version:
$ git checkout v0.12.52

• ›Adds timeout parameter to webpage readers in llama-index-readers-web, defaulting to 60 seconds.
• ›New llama-index-tools-jira-issue package (v0.1.0) introducing a Jira issue tool spec for agent use.
• ›Optimizes memory usage for BGEM3Index persistence in llama-index-indices-managed-bge-m3.

FunctionTool gains auto type conversion for basic Python types like date when using Pydantic fields.

$ git clone --branch v0.12.51 https://github.com/run-llama/llama_index.git
# already have the repo? check out this version:
$ git checkout v0.12.51

• ›Enhances FunctionTool with automatic type conversion for basic Python types (e.g., date) when declared as Pydantic fields in tool functions.

LlamaIndex v0.12.50 adds Cloudflare AI Gateway LLM, S3 vector store, ServiceNow reader, HTML table extraction, and Google Search tool support.

$ git clone --branch v0.12.50 https://github.com/run-llama/llama_index.git
# already have the repo? check out this version:
$ git checkout v0.12.50

• ›Adds google_search tool support to the llama-index-llms-google-genai GoogleGenAI LLM integration.
• ›Introduces llama-index-llms-cloudflare-ai-gateway [0.1.0], a new LLM integration for Cloudflare AI Gateway.
• ›Introduces llama-index-vector-stores-s3 [0.1.0] with S3 Vectors support as a new vector store backend.
• ›Adds llama-index-readers-service-now [0.1.0], a new reader for ServiceNow data.
• ›Adds HTML table extraction support to MarkdownElementNodeParser in llama-index-core.

LlamaIndex v0.12.49 adds structured output in agents, DuckDB stores, Moorcheh vector store, and retry for workflow agents.

$ git clone --branch v0.12.49 https://github.com/run-llama/llama_index.git
# already have the repo? check out this version:
$ git checkout v0.12.49

• ›Adds retry capability to workflow agents in llama-index-core.
• ›Adds structured output support in agents (llama-index-core) as a first implementation.
• ›Adds llama-index-storage-kvstore-duckdb [0.1.3], llama-index-storage-docstore-duckdb [0.1.0], and llama-index-storage-index-store-duckdb [0.1.0] packages, providing DuckDB-backed KV, document, and index stores.
• ›Adds async support and faster cosine similarity to llama-index-vector-stores-duckdb.
• ›Adds llama-index-vector-stores-moorcheh [0.1.0] with a new Moorcheh vector store integration.

LlamaIndex v0.12.48 adds cached content support for GoogleGenAI and image prompt support for OCI Generative AI Llama models.

$ git clone --branch v0.12.48 https://github.com/run-llama/llama_index.git
# already have the repo? check out this version:
$ git checkout v0.12.48

• ›Adds cached content support to the llama-index-llms-google-genai integration (v0.2.4), enabling reuse of cached context in GoogleGenAI LLM calls.
• ›Adds image prompt support for OCI Generative AI Llama models in llama-index-llms-oci-genai (v0.5.1).
• ›Reduces trips to the KV store during Document Hash Checks in llama-index-core, improving performance for large document ingestion workflows.

LlamaIndex v0.12.47 adds agent iteration limits, forced tool calling, Anthropic citations, LanceDB multimodal integration, and OCI GenAI image prompts.

$ git clone --branch v0.12.47 https://github.com/run-llama/llama_index.git
# already have the repo? check out this version:
$ git checkout v0.12.47

result = agent.run('Summarize the top 5 findings from this report', max_iterations=20)

• ›Adds default max_iterations argument (value: 20) to the .run() method on agents in llama-index-core, capping runaway agent loops out of the box.
• ›Sets tool_required=True by default in FunctionCallingProgram and structured LLMs where supported, ensuring tool calls are always attempted rather than optionally skipped.
• ›New Anthropic citations support in llama-index-llms-anthropic v0.7.6.
• ›Adds image prompt support for OCI Generative AI Llama models in llama-index-llms-oci-genai.
• ›New llama-index-indices-managed-lancedb v0.1.0 integration for LanceDB MultiModal AI LakeHouse.

LlamaIndex v0.12.46 adds async delete and insert methods to VectorStoreIndex.

$ git clone --branch v0.12.46 https://github.com/run-llama/llama_index.git
# already have the repo? check out this version:
$ git checkout v0.12.46

• ›Adds async delete and insert methods to VectorStoreIndex in llama-index-core, enabling non-blocking vector store mutations in async workflows.

LlamaIndex v0.12.45 adds tool content block output, chat UI events, AWS Bedrock Claude models, and async Google Search support.

$ git clone --branch v0.12.45 https://github.com/run-llama/llama_index.git
# already have the repo? check out this version:
$ git checkout v0.12.45

from llama_index.embeddings.azure_openai import AzureOpenAIEmbedding

embed_model = AzureOpenAIEmbedding(
    model="text-embedding-3-large",
    deployment_name="my-embedding-deployment",
    dimensions=512,
    azure_endpoint="https://<your-resource>.openai.azure.com/",
    api_key="<your-api-key>",
)

• ›Adds dimensions parameter to AzureOpenAIEmbedding in llama-index-embeddings-azure-openai for controlling embedding output size.
• ›Allows tools to output content blocks in llama-index-core, enabling richer structured tool responses.
• ›Adds chat UI events and models to the llama-index-core package.
• ›Adds new AWS Claude models available on Bedrock to llama-index-llms-anthropic.
• ›Adds proper async Google Search support to GoogleSearchToolSpec in llama-index-tools-google.

AutoGen 0.6.4 adds reflection for Claude in AssistantAgent, Workbench tool-name overrides, and Qwen2.5VL support.

$ git clone --branch python-v0.6.4 https://github.com/microsoft/autogen.git
# already have the repo? check out this version:
$ git checkout python-v0.6.4

result = await graph_flow.run(task="Analyze this dataset.")
print(result.stop_reason)  # termination message now lives here, not in result.messages

• ›Enables GraphFlow to resume with a new or empty task after a termination condition without an explicit reset, matching the behavior of RoundRobinGroupChat and SelectorGroupChat.
• ›Adds tool name and description override support to McpWorkbench and StaticWorkbench, allowing client-side customization of server-side tool metadata.
• ›Adds reflection support for Claude models in AssistantAgent.
• ›Adds Qwen2.5VL vision-language model support.

AutoGen v0.6.2 adds streaming tools, inner tool-call loops, OTel GenAI traces, Mem0 memory, and a tool_choice parameter.

$ git clone --branch python-v0.6.2 https://github.com/microsoft/autogen.git
# already have the repo? check out this version:
$ git checkout python-v0.6.2

from autogen_agentchat.agents import AssistantAgent
from autogen_agentchat.tools import AgentTool

sub_agent = AssistantAgent(name="sub", model_client=model_client)
tool = AgentTool(agent=sub_agent)

main_agent = AssistantAgent(name="main", model_client=model_client, tools=[tool])

async for event in main_agent.run_stream(task="Summarize the report"):
    print(event)

from autogen_agentchat.agents import AssistantAgent

agent = AssistantAgent(
    name="analyst",
    model_client=model_client,
    tools=[search_tool, calculator_tool],
    max_tool_iterations=5,
)
result = await agent.run(task="Find and compute the average price of the top 10 items.")
print(result.messages[-1].content)

from autogen_core.tools import BaseStreamTool
from typing import AsyncGenerator

class MyStreamTool(BaseStreamTool):
    async def run_stream(
        self, args: dict, cancellation_token=None
    ) -> AsyncGenerator[str, None]:
        for chunk in do_work(args["input"]):
            yield chunk

• ›Adds streaming tool support via autogen_core.tools.BaseStreamTool and autogen_core.tools.StreamWorkbench, exposing inner agent/team events through AgentTool and TeamTool when used with AssistantAgent.
• ›Adds tool_choice parameter to ChatCompletionClient create and create_stream methods for explicit tool selection control.
• ›Enables an inner tool-calling loop in AssistantAgent via the new max_tool_iterations constructor parameter, looping until the model stops generating tool calls or the limit is reached.
• ›Adds OpenTelemetry GenAI semantic-convention traces (create_agent, invoke_agent, execute_tool) for agents and tools; disable with AUTOGEN_DISABLE_RUNTIME_TRACING=true.
• ›Adds output_task_messages flag to run and run_stream to control whether input task messages are emitted in the event stream.

OpenAI Agents SDK v0.2.3 adds direct access to the model layer from a realtime session.

$ git clone --branch v0.2.3 https://github.com/openai/openai-agents-python.git
# already have the repo? check out this version:
$ git checkout v0.2.3

• ›Adds direct access to the model layer from a realtime session, enabling lower-level control over the realtime model interface.

OpenAI Agents SDK v0.2.1 adds beta Realtime agents with handoffs and MCP structuredContent support.

$ git clone --branch v0.2.1 https://github.com/openai/openai-agents-python.git
# already have the repo? check out this version:
$ git checkout v0.2.1

• ›Supports structuredContent in MCP tool_result responses, enabling richer structured data from MCP tools.
• ›Introduces Realtime agents (beta) with support for handoffs between agents during live audio/streaming sessions.
• ›Adds streaming of function call arguments to Chat Completions.

OpenAI Agents SDK v0.2.0 adds Sessions for conversation history, beta RealtimeAgent support, MCP prompts, and file_input content.

$ git clone --branch v0.2.0 https://github.com/openai/openai-agents-python.git
# already have the repo? check out this version:
$ git checkout v0.2.0

from pydantic import Field
from openai_agents import function_schema

@function_schema
def search_cve(cve_id: str = Field(..., description="CVE identifier, e.g. CVE-2024-1234"),
              severity: str = Field("high", description="Minimum severity filter")) -> str:
    ...

• ›Introduces Sessions API for automatic conversation history management, letting agents maintain context across multiple turns without manual history threading.
• ›Adds RealtimeAgent class (beta) with a dedicated RealtimeSession, OpenAI realtime transport implementation, guardrail support, and built-in tracing.
• ›Adds on_start support to VoiceWorkflowBase and VoicePipeline for lifecycle hooks at session start.
• ›Supports file_input content type in agent inputs.
• ›Supports MCP prompts via the MCP integration layer.

PydanticAI v0.4.8 adds tenacity retry integration and thinking-part tracing in OpenTelemetry model response events.

$ git clone --branch v0.4.8 https://github.com/pydantic/pydantic-ai.git
# already have the repo? check out this version:
$ git checkout v0.4.8

• ›Adds tenacity utilities and integration for improved retry handling in agent workflows.
• ›Includes ThinkingPart in OpenTelemetry OTEL events emitted via ModelResponse, surfacing model reasoning in traces.

PydanticAI v0.4.7 adds MoonshotAI, Vercel AI Gateway providers, Gemini Files API support, and MCP ResourceLink handling.

$ git clone --branch v0.4.7 https://github.com/pydantic/pydantic-ai.git
# already have the repo? check out this version:
$ git checkout v0.4.7

from pydantic_ai.mcp import MCPServer

server = MCPServer(
    url='https://mcp.example.com/sse',
    read_timeout=30,
)

• ›Renames MCPServer parameter sse_read_timeout to read_timeout, which is now passed through to ClientSession.
• ›Adds MoonshotAI provider with Kimi-K2 model support.
• ›Adds Vercel AI Gateway provider.
• ›Supports passing files uploaded to the Gemini Files API and setting a custom media type.
• ›Parses <think> tags in streamed text as thinking parts (ThinkingPart).

PydanticAI v0.4.6 adds URL and binary PDF support for the Mistral provider.

$ git clone --branch v0.4.6 https://github.com/pydantic/pydantic-ai.git
# already have the repo? check out this version:
$ git checkout v0.4.6

• ›Adds URL and binary PDF input support for the Mistral provider, enabling document-based prompts via URL or raw binary PDF.
• ›Speeds up the internal _estimate_string_tokens function, improving throughput for token-heavy workloads.

PydanticAI v0.4.5 adds streamable HTTP transport support to mcp-run-python and changes format_as_xml defaults.

$ git clone --branch v0.4.5 https://github.com/pydantic/pydantic-ai.git
# already have the repo? check out this version:
$ git checkout v0.4.5

• ›Supports streamable HTTP transport in mcp-run-python, enabling streaming MCP server connections over HTTP.

PydanticAI v0.4.4 adds Toolsets, AG-UI protocol support, new OpenAI/Grok/Kimi models, and an identifier field on BinaryContent.

$ git clone --branch v0.4.4 https://github.com/pydantic/pydantic-ai.git
# already have the repo? check out this version:
$ git checkout v0.4.4

from pydantic_ai.messages import BinaryContent

image = BinaryContent(data=image_bytes, media_type='image/png', identifier='screenshot-001')
result = await agent.run([image, 'Describe this image.'])

• ›Adds identifier field to the BinaryContent class for tagging binary content objects.
• ›Introduces Toolsets and Deferred Tools, enabling grouped and lazily-resolved tool registration on agents.
• ›Supports the AG-UI protocol for frontend-agent communication.
• ›Adds OpenAI models o1-pro, o3-pro, o3-deep-research, and computer-use as selectable models.
• ›Adds grok-4 and kimi-k2 (via Groq) as selectable models.

PydanticAI v0.4.3 adds Hugging Face provider support, output function tracing, and base64 encoding for tool returns.

$ git clone --branch v0.4.3 https://github.com/pydantic/pydantic-ai.git
# already have the repo? check out this version:
$ git checkout v0.4.3

• ›Adds base64 encoding support to tool_return_ta, enabling binary data to be returned from tools.
• ›Adds output function tracing, allowing agent output functions to be captured in traces.
• ›Adds Hugging Face as a new model provider.

PydanticAI v0.4.2 adds StructuredDict for custom JSON schema outputs, model settings on model classes, and DeepSeek reasoning_content streaming support.

$ git clone --branch v0.4.2 https://github.com/pydantic/pydantic-ai.git
# already have the repo? check out this version:
$ git checkout v0.4.2

• ›Adds StructuredDict class for defining structured outputs with a custom JSON schema, giving callers direct control over the schema shape returned by the model.
• ›Allows model settings to be passed directly to model classes, enabling per-model configuration at instantiation time.
• ›Supports DeepSeek reasoning_content field in streamed responses, surfacing chain-of-thought reasoning tokens from DeepSeek models during streaming.
• ›Speeds up internal _ensure_decodeable function by 634%, unlocking higher-throughput decoding for workloads processing large volumes of model output.

PydanticAI v0.4.1 adds sync task evaluation support and drops FastA2A as a transitive dependency.

$ git clone --branch v0.4.1 https://github.com/pydantic/pydantic-ai.git
# already have the repo? check out this version:
$ git checkout v0.4.1

• ›Adds support for evaluating synchronous tasks in PydanticAI's evals framework, expanding coverage beyond async-only workflows.

PydanticAI v0.4.0 adds broader Gemini audio support and makes ToolDefinition.description optional.

$ git clone --branch v0.4.0 https://github.com/pydantic/pydantic-ai.git
# already have the repo? check out this version:
$ git checkout v0.4.0

• ›Makes ToolDefinition.description optional, removing the requirement to supply a description when defining tools.
• ›Adds all Gemini-supported audio types to AudioUrl, expanding multimodal input coverage for Gemini models.
• ›Retains default values in non-strict OpenAI schemas, preserving schema fidelity when targeting OpenAI backends.

PydanticAI v0.3.7 adds GitHub Models provider, ACI.dev Tools integration, sync streaming, and Google video analysis args.

$ git clone --branch v0.3.7 https://github.com/pydantic/pydantic-ai.git
# already have the repo? check out this version:
$ git checkout v0.3.7

• ›Adds model_request_stream_sync to the direct API, enabling synchronous streaming of model requests.
• ›Adds GitHub Models as a new provider via the GitHub Models provider integration.
• ›Adds support for Google-specific arguments for video analysis in the Google provider.
• ›Implements ACI.dev Tools integration, providing a convenient way to use ACI.dev tools in PydanticAI.
• ›AgentStream.stream_output (available inside agent.iter) now streams validated output data instead of raising validation errors mid-stream.

PydanticAI v0.3.6 adds predicted outputs to OpenAIModelSettings and records tool responses in trace spans.

$ git clone --branch v0.3.6 https://github.com/pydantic/pydantic-ai.git
# already have the repo? check out this version:
$ git checkout v0.3.6

from pydantic_ai.models.openai import OpenAIModelSettings

settings = OpenAIModelSettings(
    predicted_outputs={"type": "content", "content": "<your predicted text here>"}
)
result = await agent.run("Refactor this code", model_settings=settings)

• ›Adds support for predicted_outputs in OpenAIModelSettings, enabling speculative/predicted output hints when calling OpenAI models.
• ›Records tool response data in tool-run spans, enriching tracing and observability for agent tool calls.
• ›Improves model communication by marking a RetryPromptPart not tied to a tool call as validation feedback rather than a user message, giving the model clearer signal on why a retry is occurring.
• ›Switches agent overriding from a local attribute to contextvars, making agent context propagation safe across async/concurrent workloads.

Semantic Kernel Python 1.35.0 adds gpt-image-1 support and partial result emission for the magentic orchestration pattern.

$ git clone --branch python-1.35.0 https://github.com/microsoft/semantic-kernel.git
# already have the repo? check out this version:
$ git checkout python-1.35.0

• ›Adds support for gpt-image-1 model integration.
• ›Emits partial results for the magentic orchestration pattern when retrieving the final result, if one is available.
• ›Introduces message cache usage in agent orchestrations to improve efficiency.
• ›Improves exception handling in orchestration flows.

Semantic Kernel .NET 1.60.0 adds Retrieval API plugin, SK-to-MEAI content converters, ChatSystem/DeveloperPrompt support, and promotes AI connectors out of experimental.

$ git clone --branch dotnet-1.60.0 https://github.com/microsoft/semantic-kernel.git
# already have the repo? check out this version:
$ git checkout dotnet-1.60.0

using Microsoft.SemanticKernel;

ChatMessageContent skContent = new(AuthorRole.Assistant, "Hello!");
var meaiContent = skContent.ToAIContent();

• ›Adds ChatSystem and DeveloperPrompt properties to AzureOpenAIPromptExecutionSettings and OpenAIPromptExecutionSettings for ChatClients, enabling system and developer prompt injection at the settings level.
• ›Adds Filter support to TextSearchProvider, allowing callers to narrow text search results programmatically.
• ›Exposes conversion helpers from SK content types to Microsoft.Extensions.AI (MEAI) content primitives, bridging SK's ChatMessageContent and related types to MEAI's content model.
• ›Adds a Retrieval API Plugin to CAPs (Copilot Agent Plugins), surfacing retrieval as a first-class plugin capability.
• ›Removes the SKEXP0070 experimental attribute from non-GA AI connectors, graduating them to stable API surface.

Semantic Kernel vectordata-dotnet-9.7.0 adds a Retrieval API plugin, SK-to-MEAI content conversion helpers, A2A agent support, ONNX ChatClient extensions, and more.

$ git clone --branch vectordata-dotnet-9.7.0 https://github.com/microsoft/semantic-kernel.git
# already have the repo? check out this version:
$ git checkout vectordata-dotnet-9.7.0

merged_args = KernelArguments(foo='bar') | KernelArguments(baz='qux')
result = await kernel.invoke(my_function, merged_args)

• ›Adds Filter support to TextSearchProvider for scoped vector text search queries.
• ›Exposes conversion helpers from SK Contents to MEAI (Microsoft.Extensions.AI) content primitives, easing interop between SK and MEAI pipelines.
• ›Adds ChatSystem/DeveloperPrompt support to {Azure}OpenAIPromptExecutionSettings for ChatClient-based usage.
• ›Adds ONNX ChatClient extensions, enabling ONNX-backed models to be used via the ChatClient abstraction.
• ›Adds the Retrieval API Plugin to Conversational AI Primitives (CAPs).

Semantic Kernel .NET 1.59.0 adds web/file search sample, exposes GeminiKernelFunctionMetadataExtensions, and lets clients remove the safe_prompt attribute from JSON.

$ git clone --branch dotnet-1.59.0 https://github.com/microsoft/semantic-kernel.git
# already have the repo? check out this version:
$ git checkout dotnet-1.59.0

using Microsoft.SemanticKernel.Connectors.Google;

var metadata = GeminiKernelFunctionMetadataExtensions.ToGeminiFunctionMetadata(function.Metadata);

• ›Exposes GeminiKernelFunctionMetadataExtensions publicly, making Gemini function metadata utilities available to library consumers.
• ›Enables clients to remove the safe_prompt attribute from JSON in Mistral/compatible connector requests.
• ›Adds a sample demonstrating how to use web and file search together with Semantic Kernel agents.
• ›Ignores unknown response item types instead of throwing, improving forward-compatibility with evolving model response schemas.

browser-use 0.5.6 adds CDP URL telemetry, ARIA menu dropdown support, typed package marker, and speed improvements.

$ git clone --branch 0.5.6 https://github.com/browser-use/browser-use.git
# already have the repo? check out this version:
$ git checkout 0.5.6

• ›Adds py.typed marker to the package, enabling type-checker support (mypy, pyright) for downstream consumers.
• ›Adds CDP (Chrome DevTools Protocol) URL to Agent Telemetry events for richer session observability.
• ›Adds ARIA menu support to dropdown interaction functions, broadening the range of web UI components the agent can operate.
• ›Disables screenshot capture automatically when vision is disabled, reducing unnecessary overhead.
• ›Handles PDF viewer content via the read-file action, allowing agents to extract text from in-browser PDF viewers.

browser-use 0.5.5 adds Flash Mode, DeepSeek and Groq support, and OpenAI CUA fallback

$ git clone --branch 0.5.5 https://github.com/browser-use/browser-use.git
# already have the repo? check out this version:
$ git checkout 0.5.5

• ›Adds DeepSeek LLM chat model integration as a supported provider.
• ›Adds tool-calling support for Groq-hosted models.
• ›Adds OpenAI CUA (Computer-Using Agent) fallback mode for browser automation.
• ›Introduces Flash Mode for faster browser-use agent operation.
• ›Limits the wait action to a maximum of 10 seconds, capping runaway waits.

browser-use 0.5.4 adds automatic agent crash recovery and consolidates BrowserSession navigation methods

$ git clone --branch 0.5.4 https://github.com/browser-use/browser-use.git
# already have the repo? check out this version:
$ git checkout 0.5.4

await session.navigate(url='https://example.com', new_tab=True, timeout_ms=15000)

• ›Adds @require_healthy_browser(usable_page=True, reopen_page=True) decorator in browser_use/browser/session.py for crash-resilient browser operations.
• ›Combines navigate(), navigate_to(), create_new_tab(), new_page() and other redundant BrowserSession helper methods into a single navigate(url: str, new_tab: bool, timeout_ms: int) method.
• ›Agent now auto-recovers from crashed or stalled pages: retries the stalled page via JS page.evaluate(1), reopens the URL in a new tab, retreats to about:blank, relaunches a crashed browser with original settings, and falls back to a tmp incognito user_data_dir=None (with storage_state.json cookies) if the browser fails to relaunch.
• ›Adds PDF file creation support in the agent's file-handling actions.
• ›Exposes retry decorator @retry(timeout=5, wait=1, retries=2, ...) from bubus/helpers.py for use in custom actions.

browser-use 0.5.3 adds automatic PDF downloads and graceful incognito fallback for unusable user_data_dir profiles

$ git clone --branch 0.5.3 https://github.com/browser-use/browser-use.git
# already have the repo? check out this version:
$ git checkout 0.5.3

from browser_use import BrowserSession

session = BrowserSession(user_data_dir='/home/user/.config/chrome-profile')
# If the profile dir is unusable, falls back to user_data_dir=None automatically

• ›Adds graceful fallback to a temporary incognito profile (user_data_dir=None) when BrowserSession(user_data_dir='/path/to/some/profile') fails to launch due to corruption, SingletonLock conflicts, or filesystem permission issues — instead of crashing.
• ›Automatically downloads PDFs when the browser navigates to one, with scrolling inside PDFs via pure CDP.
• ›Takes base64 CDP screenshots directly without going through Playwright, enabling faster screen capture.

browser-use 0.5.0 adds native bidirectional MCP support, exposing external MCP tools to the agent and the agent itself as an MCP server.

$ git clone --branch 0.5.0 https://github.com/browser-use/browser-use.git
# already have the repo? check out this version:
$ git checkout 0.5.0

{
  "mcpServers": {
    "browser-use": {
      "command": "uvx",
      "args": ["browser-use[cli]", "--mcp"]
    }
  }
}

• ›Adds --mcp CLI flag (via browser-use[cli]) to launch the Browser Use agent as an MCP server callable by any MCP client, including Claude Desktop.
• ›Adds MCP client support so external MCP servers and their tools can be connected to the Browser Use agent and used as actions.
• ›Expands ~/.config/browseruse/config.json schema with new fields for MCP client and server connectors.
• ›Supports installing Browser Use as a Claude Desktop extension via a browser-use.dxt file or manual entry in the Claude Desktop mcpServers config block.
• ›Enhances scroll actions with pixel-level control.

browser-use 0.4.5 adds OpenRouter support, JSON/CSV/PDF data extraction, Gmail OTP integration, and multi-image-per-step agent input.

$ git clone --branch 0.4.5 https://github.com/browser-use/browser-use.git
# already have the repo? check out this version:
$ git checkout 0.4.5

from langchain_openai import ChatOpenAI

llm = ChatOpenAI(
    base_url='https://openrouter.ai/api/v1',
    api_key='<your-openrouter-key>',
    model='openai/gpt-4o',
)
agent = Agent(task='<task>', llm=llm)

• ›Adds BrowserSettings to BrowserProfile for configuring browser-level settings directly on the profile object.
• ›Supports reading and extracting structured data from JSON, CSV, and PDF files as agent actions.
• ›Integrates the Gmail API to retrieve OTPs and email content during automated workflows.
• ›Adds OpenRouter as a supported LLM provider for driving agents.
• ›Enables multiple screenshots per agent step, each with a label, as LLM input — improving visual context for multi-action steps.

Cline v3.20.2 adds Git Bash terminal support on Windows.

$ git clone --branch v3.20.2 https://github.com/cline/cline.git
# already have the repo? check out this version:
$ git checkout v3.20.2

• ›Adds Git Bash terminal support for Windows users, with accompanying documentation for resolving Windows terminal issues.

Cline v3.20.0 adds Qwen 3 model support, Devtral Medium on Mistral, and credit balance visibility for all accounts.

$ git clone --branch v3.20.0 https://github.com/cline/cline.git
# already have the repo? check out this version:
$ git checkout v3.20.0

• ›Adds support for Qwen 3 models on the Qwen provider.
• ›Adds Devtral Medium model to the Mistral provider.
• ›Displays credit balance for all accounts in the account view.

Cline v3.19.8 adds separate Plan/Act model settings, automated release announcements, and updated Cerebras models.

$ git clone --branch v3.19.8 https://github.com/cline/cline.git
# already have the repo? check out this version:
$ git checkout v3.19.8

• ›Adds separate model settings for Plan and Act modes, letting users assign different models to each phase.
• ›Automates announcement display for major.minor releases within the UI.
• ›Updates available Cerebras models and modifies their context window sizes.
• ›Uses --system-prompt-file to pass the system prompt to Claude Code.

Cline v3.19.7 adds Hugging Face as a new AI provider and introduces SAP AI Core documentation.

$ git clone --branch v3.19.7 https://github.com/cline/cline.git
# already have the repo? check out this version:
$ git checkout v3.19.7

• ›Adds Hugging Face as a supported AI provider.
• ›Introduces SAP AI Core documentation for enterprise provider setup.

Cline v3.19.5 adds Groq provider support, vision for Moonshot v1, and org role/credit visibility.

$ git clone --branch v3.19.5 https://github.com/cline/cline.git
# already have the repo? check out this version:
$ git checkout v3.19.5

• ›Adds Groq as a supported AI provider.
• ›Adds vision capability to the Moonshot v1 model.
• ›Displays user role within an organization in the UI.
• ›Shows credit purchase link contextually based on the active organization.

Cline v3.19.4 adds a Chinese endpoint option for the Moonshot provider.

$ git clone --branch v3.19.4 https://github.com/cline/cline.git
# already have the repo? check out this version:
$ git checkout v3.19.4

• ›Adds ability to select a Chinese endpoint when configuring the Moonshot provider.

Cline v3.19.3 adds Moonshot AI as a supported LLM provider.

$ git clone --branch v3.19.3 https://github.com/cline/cline.git
# already have the repo? check out this version:
$ git checkout v3.19.3

• ›Adds Moonshot AI as a new LLM provider option.

Cline API errors now include a request ID for faster incident triage.

$ git clone --branch v3.19.2 https://github.com/cline/cline.git
# already have the repo? check out this version:
$ git checkout v3.19.2

• ›Adds request ID to errors returned by the Cline API, enabling precise correlation of failures to specific requests.

Cline v3.18.15 adds Bedrock API Key auth, kimi-k2 model support, Groq provider, and markdown rendering in MCP responses.

$ git clone --branch v3.18.15 https://github.com/cline/cline.git
# already have the repo? check out this version:
$ git checkout v3.18.15

• ›Supports Amazon Bedrock API Key authentication as a new auth method alongside existing IAM credential flows.
• ›Adds kimi-k2 as a trending model option with Together and Groq as available providers.
• ›Renders markdown formatting in MCP tool responses for improved readability.
• ›Introduces DiffService and platform-specific DiffViewProvider to the host bridge, enabling diff editor integration on both VS Code and external platforms.

Cline v3.18.13 adds a git branch analysis workflow and a re-sign-in button to the account view.

$ git clone --branch v3.18.13 https://github.com/cline/cline.git
# already have the repo? check out this version:
$ git checkout v3.18.13

• ›Adds a git branch analysis workflow for reviewing branch-level changes.
• ›Adds a re-sign-in button to the account UI when authentication needs to be refreshed.

Cline v3.18.10 adds Grok 4 support and Gemini 2.5 Flash Preview with thinking token config for Gemini 2.5 Pro.

$ git clone --branch v3.18.10 https://github.com/cline/cline.git
# already have the repo? check out this version:
$ git checkout v3.18.10

• ›Adds Grok 4 model support (recommended provider option).
• ›Adds Gemini 2.5 Flash Preview model support.
• ›Adds thinking token configuration for Gemini 2.5 Pro.

Cline v3.18.6 adds SAP AI Core support and org/personal inference switching with usage/credit reporting.

$ git clone --branch v3.18.6 https://github.com/cline/cline.git
# already have the repo? check out this version:
$ git checkout v3.18.6

• ›Adds tracking header support for SAP AI Core as a provider integration.
• ›Enables organization and personal inference switching, account switching, and usage/credit reporting within Cline.

Cline v3.18.5 persists plan/act mode globally across sessions and improves provider-switching performance.

$ git clone --branch v3.18.5 https://github.com/cline/cline.git
# already have the repo? check out this version:
$ git checkout v3.18.5

• ›Globally persists plan/act mode across sessions so your preferred mode is remembered after restarts.
• ›Persists chat settings (e.g. language) at the workspace level for consistent per-project configuration.
• ›Optimizes provider switching performance with batched storage operations.

Cline v3.18.4 adds Gemini 2.5 Pro and Flash support for SAP AI Core users.

$ git clone --branch v3.18.4 https://github.com/cline/cline.git
# already have the repo? check out this version:
$ git checkout v3.18.4

• ›Adds Gemini 2.5 Pro and Flash model support to the SAP AI Core provider.

Cline v3.18.2 adds Claude Sonnet 4/Opus 4 via SAP AI Core, litellm session grouping, and Thinking Budget for Claude Code.

$ git clone --branch v3.18.2 https://github.com/cline/cline.git
# already have the repo? check out this version:
$ git checkout v3.18.2

• ›Adds confirmation popup when deleting tasks to prevent accidental data loss.
• ›Supports Claude Sonnet 4 and Opus 4 models in the SAP AI Core provider.
• ›Supports litellm_session_id to group multiple requests into a single session.
• ›Adds 'Thinking Budget' customization option for Claude Code.

Adds a Vertex AI OpenAI adapter and IntelliJ terminal integration including runCommand() and getTerminalContents().

$ 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 slug field to rule objects in dev data.
• ›New Vertex AI OpenAI adapter enables routing Continue through Google Cloud Vertex AI endpoints.
• ›Implements runCommand() method in IntelliJIde.kt with TerminalOptions support, aligning IntelliJ terminal behavior with the VS Code extension.
• ›Adds getTerminalContents() in IntelliJIde.kt to read content from the selected terminal panel.
• ›Terminal panels now reuse existing terminals by matching terminalName before spawning a new one.

Continue config-yaml 1.3.0 adds slug field to rule objects in dev data.

$ 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 slug field to rule objects in dev data for identifying rules by a stable, human-readable identifier.

Continue v1.0.19 adds Plan Mode, Amazon Nova/claude-opus-4/LlamaStack support, VertexAI key auth, MCP cwd, and mermaid diagram rendering.

$ git clone --branch v1.0.19-vscode https://github.com/continuedev/continue.git
# already have the repo? check out this version:
$ git checkout v1.0.19-vscode

mcpServers:
  - name: my-mcp-server
    command: npx
    args: ['-y', 'my-mcp-package']
    cwd: /workspace/my-project

models:
  - provider: vertexai
    model: gemini-2.0-flash
    keyFile: /secrets/gcp-sa-key.json
    project: my-gcp-project
    region: us-central1

• ›Adds cwd option to MCP server configuration in the Continue YAML config, allowing working-directory control for MCP tools.
• ›Adds keyFile, keyJson, and express/apiKey fields to VertexAI provider configuration for flexible credential management.
• ›Adds uri field for context items in the HTTP context provider.
• ›Adds tool call support for deepseek-chat and deepseek-reasoner models.
• ›Adds claude-opus-4 tool support for Amazon Bedrock models.

Continue v1.0.18 adds regex rule triggers, hot-reloading rules, vLLM reranking, Ollama tool-call streaming, and a search-and-replace tool experiment.

$ git clone --branch v1.0.18-vscode https://github.com/continuedev/continue.git
# already have the repo? check out this version:
$ git checkout v1.0.18-vscode

# .continue/rules/no-use-effect.md
---
name: No useEffect
trigger:
  regex: "\\.tsx?$"
---
Do not use useEffect. Prefer derived state or event handlers.

mcpServers:
  - name: my-mcp
    command: npx
    args: ['-y', 'my-mcp-server']
    requestOptions:
      timeout: 30000
      headers:
        Authorization: 'Bearer <token>'

models:
  - name: my-model
    provider: ollama
    model: llama3
    minP: 0.05
    frequencyPenalty: 0.3
    presencePenalty: 0.3

• ›Adds requestOptions to YAML config for MCP server definitions.
• ›Adds minP, frequencyPenalty, and presencePenalty parameters to YAML config model blocks.
• ›Adds proxy support for ripgrep download, enabling use in air-gapped or proxy-restricted environments.
• ›Supports regex-based rule triggers in .continue/rules files, letting rules activate automatically when file content or context matches a pattern.
• ›Adds manual on/off toggling of individual rules during a session.

Continue v1.0.16 adds vLLM reranking, regex rule triggers, hot-reload for rules, Ollama tool-call streaming, and a search-and-replace tool experiment.

$ git clone --branch v1.0.16-vscode https://github.com/continuedev/continue.git
# already have the repo? check out this version:
$ git checkout v1.0.16-vscode

mcpServers:
  my-server:
    command: npx
    args: ['-y', 'my-mcp-server']
    requestOptions:
      timeout: 30000
      headers:
        Authorization: 'Bearer <token>'

models:
  - name: my-model
    provider: ollama
    model: llama3
    minP: 0.05
    frequencyPenalty: 0.3
    presencePenalty: 0.2

• ›Adds requestOptions to YAML config for MCP servers, enabling per-server HTTP options in .continue/config.yaml.
• ›YAML config now loads minP, frequencyPenalty, and presencePenalty model parameters.
• ›Adds proxy support for ripgrep download via environment-level fetch configuration.
• ›Adds CONTINUE_CUSTOM_FETCH_ENV flag (Custom Fetch Env Flag) to control fetch behaviour via environment variable.
• ›Reads NODE_EXTRA_CA_CERTS and improves NO_PROXY logic in the core fetch layer.

Goose v1.1.0 ships a redesigned desktop UI with sidebar and settings tabs, new recipe list command, glob/grep file search tools, and Windows CLI installer.

$ git clone --branch v1.1.0 https://github.com/block/goose.git
# already have the repo? check out this version:
$ git checkout v1.1.0

$ goose recipe list

$ /clear

• ›Adds recipe list subcommand to enumerate available recipes from the CLI.
• ›New /clear command in CLI sessions clears the persisted session file.
• ›Adds glob search and grep tools to improve file search capabilities.
• ›Implements OpenAI streaming support for faster, real-time LLM responses.
• ›Complete redesign of the Goose desktop UI with a sidebar and settings tabs.

Goose v1.0.36 adds streaming LLM responses, cost estimation, OpenRouter model discovery, vi edit mode, and new provider/model CLI flags.

$ git clone --branch v1.0.36 https://github.com/block/goose.git
# already have the repo? check out this version:
$ git checkout v1.0.36

$ goose run --provider openai --model gpt-4o <session-or-prompt>

• ›Streams LLM responses in real time instead of waiting for full completion.
• ›Adds per-provider cost estimation in the Goose CLI so practitioners can compare spend before committing to a provider.
• ›Fetches OpenRouter's supported model list dynamically inside goose configure.
• ›Exposes AZURE_OPENAI_API_KEY as a visible, configurable parameter in the config system.
• ›Adds --provider and --model CLI options to the goose run command for one-off provider/model selection.

Goose v1.0.32 adds OAuth 2.0 for MCP, streamable-HTTP transport, JSON schema recipe validation, and sub-recipes from GitHub.

$ git clone --branch v1.0.32 https://github.com/block/goose.git
# already have the repo? check out this version:
$ git checkout v1.0.32

$ goose recipe validate <path-to-recipe.yaml>

• ›Adds fuzzy file search functionality for navigating files.
• ›Adds JSON schema validation to the goose recipe validate CLI command.
• ›Introduces max_turns setting for the agent to cap autonomous runs without user input.
• ›Enables structured output support in recipes, usable from both CLI and GUI.
• ›GUI now renders structured output defined in recipes.

Goose v1.0.31 adds sub-recipe chaining from the CLI, async token counting, and Claude 4 model support.

$ git clone --branch v1.0.31 https://github.com/block/goose.git
# already have the repo? check out this version:
$ git checkout v1.0.31

• ›Supports passing additional sub-recipes via the command line for composable recipe workflows.
• ›Adds Claude 4 models as available provider options.
• ›Implements async token counter with network resilience and performance optimizations.
• ›Allows use of /dev/null for no-session mode, enabling fully stateless runs.

Codex CLI v0.8.0 adds response streaming in TUI and codex exec, plus a --json flag for JSONL output.

$ git clone --branch rust-v0.8.0 https://github.com/openai/codex.git
# already have the repo? check out this version:
$ git checkout rust-v0.8.0

$ codex exec --json 'summarize the security findings in REPORT.md'

• ›Streams model responses in real time in the TUI and when using codex exec.
• ›Adds --json flag to codex exec to print output as JSONL to stdout.
• ›Reorganizes ~/.codex/sessions into YYYY/MM/DD subfolders for easier navigation and better filesystem performance.
• ›ctrl-d now only exits the TUI when the composer is empty, preventing accidental exits mid-session.

OpenAI Codex CLI v0.6.0 adds paste summarization and an experimental codex apply command for Codex Web.

$ git clone --branch rust-v0.6.0 https://github.com/openai/codex.git
# already have the repo? check out this version:
$ git checkout rust-v0.6.0

$ codex apply

• ›New codex apply command (experimental) lets you interact with Codex Web directly from the CLI.
• ›Adds paste summarization for large pastes, automatically condensing oversized clipboard input.

Codex CLI rust-v0.5.0 adds reasoning-summary config control and Android platform support via npm.

$ git clone --branch rust-v0.5.0 https://github.com/openai/codex.git
# already have the repo? check out this version:
$ git checkout rust-v0.5.0

# In your Codex config file:
model_supports_reasoning_summaries: true

• ›New model_supports_reasoning_summaries config option to control reasoning summary behavior per model.
• ›Running Codex installed via npm on Android (process.platform === "android") now routes to the Rust CLI.

Codex CLI v0.4.0 adds shell completion generation, per-profile reasoning controls, and custom base URL support.

$ git clone --branch rust-v0.4.0 https://github.com/openai/codex.git
# already have the repo? check out this version:
$ git checkout rust-v0.4.0

$ codex completion bash

• ›Supports OPENAI_BASE_URL environment variable for the built-in openai model provider, enabling use of compatible API endpoints.
• ›Adds model_reasoning_effort and model_reasoning_summary fields to profile definitions for fine-grained reasoning control.
• ›New completion subcommand enables shell tab-completion generation for package managers like Homebrew.

Codex CLI gains a --sandbox flag and custom HTTP header support for model provider requests.

$ git clone --branch rust-v0.3.0 https://github.com/openai/codex.git
# already have the repo? check out this version:
$ git checkout rust-v0.3.0

$ codex --sandbox <sandbox-policy> "refactor the auth module"

• ›Adds --sandbox flag to control sandbox behavior from the command line.
• ›Enables configuration of custom HTTP headers when making requests to model providers.

OpenCode v0.3.90 adds TreeSitter bash parsing, per-directory markdown config, and Azure Responses API support.

$ git clone --branch v0.3.90 https://github.com/sst/opencode.git
# already have the repo? check out this version:
$ git checkout v0.3.90

• ›Uses TreeSitter to parse bash commands and detect commands that escape the current working directory.
• ›Adds per-directory markdown configuration loading for mode settings.
• ›Supports the Responses API for Azure AI backends.
• ›Adds http-referer header support for Vercel AI Gateway requests.

OpenCode v0.3.82 adds custom config file path via env var, git branch in status bar, and git diff in reverted messages.

$ git clone --branch v0.3.82 https://github.com/sst/opencode.git
# already have the repo? check out this version:
$ git checkout v0.3.82

$ OPENCODE_CONFIG=~/projects/myapp/.opencode.json opencode

• ›New OPENCODE_CONFIG environment variable lets you specify a custom config file path at startup.
• ›Displays the current git branch in the status bar with responsive layout.
• ›Shows git diff inline in reverted messages for easier review of undone changes.

OpenCode v0.3.81 adds OPENCODE_CONFIG env var for custom config file paths and strips todo tool instructions from non-Anthropic models.

$ git clone --branch v0.3.81 https://github.com/sst/opencode.git
# already have the repo? check out this version:
$ git checkout v0.3.81

$ OPENCODE_CONFIG=/path/to/my-project.json opencode

• ›Adds OPENCODE_CONFIG environment variable to specify a custom config file path, enabling per-project or per-environment configurations.
• ›Strips todo tool instructions from non-Anthropic model prompts, improving compatibility and reducing noise when using alternative LLM backends.

OpenCode v0.3.80 adds a VS Code status bar button and new terminal keybindings for faster access.

$ git clone --branch v0.3.80 https://github.com/sst/opencode.git
# already have the repo? check out this version:
$ git checkout v0.3.80

• ›Shows an OpenCode button in the VS Code status bar when the terminal is focused.
• ›Brings an existing OpenCode terminal to the front in VS Code instead of opening a duplicate.
• ›Adds Cmd+Shift+Esc keybinding in VS Code to quickly open or focus the OpenCode terminal.

OpenCode v0.3.65 adds a configurable TUI scroll speed setting and improves Windows compatibility via zip.js.

$ git clone --branch v0.3.65 https://github.com/sst/opencode.git
# already have the repo? check out this version:
$ git checkout v0.3.65

• ›Adds a scroll speed setting in the TUI for configurable navigation pace.
• ›Replaces unzip with @zip.js/zip.js for Windows compatibility.

OpenCode v0.3.61 adds mode passing into the task tool and a new Aura theme.

$ git clone --branch v0.3.61 https://github.com/sst/opencode.git
# already have the repo? check out this version:
$ git checkout v0.3.61

• ›Passes the current mode into the task tool, enabling mode-aware task execution.
• ›Adds the Aura theme as a new UI color scheme option.

OpenCode v0.3.26 collapses the session header to a single line when sharing is disabled.

$ git clone --branch v0.3.26 https://github.com/sst/opencode.git
# already have the repo? check out this version:
$ git checkout v0.3.26

• ›Collapses the session header into a single line in the TUI when sharing is disabled, reducing UI clutter.

OpenCode v0.3.24 adds Vercel AI Gateway support and Gemini tool schema sanitization.

$ git clone --branch v0.3.24 https://github.com/sst/opencode.git
# already have the repo? check out this version:
$ git checkout v0.3.24

• ›Adds Vercel AI Gateway as a supported provider integration.
• ›Adds Gemini tool schema sanitization to improve compatibility with Gemini-based provider calls.

OpenCode v0.3.18 adds AWS bearer token auth for Bedrock, MCP header support, and cleaner sharing UX.

$ git clone --branch v0.3.18 https://github.com/sst/opencode.git
# already have the repo? check out this version:
$ git checkout v0.3.18

• ›Supports AWS_BEARER_TOKEN_BEDROCK environment variable for Amazon Bedrock provider autoloading.
• ›Allows MCP servers to include custom headers in their configuration.
• ›Highlights the current session in the sessions modal for easier navigation.
• ›Removes share commands from the help menu when sharing is disabled.
• ›Removes sharing info from the session header when sharing is disabled.

OpenCode v0.3.15 adds Shift+Tab backward mode cycling, full-width layout config, enhanced private npm registry support, and copy-last-message shortcut.

$ git clone --branch v0.3.15 https://github.com/sst/opencode.git
# already have the repo? check out this version:
$ git checkout v0.3.15

• ›Adds Shift+Tab keybinding to cycle through modes in reverse order.
• ›New layout config option to render the TUI at full width.
• ›Adds a 'copy last message' action in the messages component.
• ›Enhances private npm registry support.

OpenCode v0.3.12 adds /export command, Gemini CLI prompt support, and Anthropic custom mode spoofing.

$ git clone --branch v0.3.12 https://github.com/sst/opencode.git
# already have the repo? check out this version:
$ git checkout v0.3.12

$ /export

• ›New /export command exports the current conversation to your editor.
• ›Adds support for modified Gemini CLI system prompt.
• ›Supports Anthropic with custom modes via spoof prompt.
• ›Uses a small model dedicated to conversation title generation.

OpenCode v0.3.10 adds a keymap to remove entries from recently used models and improves tool call rendering.

$ git clone --branch v0.3.10 https://github.com/sst/opencode.git
# already have the repo? check out this version:
$ git checkout v0.3.10

• ›Adds a keymap in the TUI to remove entries from the recently used models list.
• ›Reduces the number of 'Unknown' entries in tool call rendering during opencode run.
• ›Adds a dedicated small model for session title generation.

OpenCode v0.3.7 adds Anthropic Console login, job-control suspend, and restores the task tool.

$ git clone --branch v0.3.7 https://github.com/sst/opencode.git
# already have the repo? check out this version:
$ git checkout v0.3.7

• ›Supports Anthropic Console login flow for direct browser-based authentication.
• ›Adds job-control suspend support (Ctrl+Z / SIGTSTP) to pause and background the process like a native shell command.

OpenCode v0.3.3 adds word-jump key bindings and new-session creation from the session dialog.

$ git clone --branch v0.3.3 https://github.com/sst/opencode.git
# already have the repo? check out this version:
$ git checkout v0.3.3

• ›Adds Ctrl+Left / Ctrl+Right arrow key support for word-by-word cursor navigation in the TUI.
• ›Enables creating a new session directly from the session dialog without leaving the UI.

OpenCode v0.3.2 adds configurable sharing options and expands file-listing ignore patterns.

$ git clone --branch v0.3.2 https://github.com/sst/opencode.git
# already have the repo? check out this version:
$ git checkout v0.3.2

• ›Adds auto/disabled sharing config options for controlling session share behavior.
• ›Expands ignore patterns in the ls tool to reduce noise from common non-source files.

OpenCode v0.2.28 adds a new API endpoint for retrieving GitHub App tokens.

$ git clone --branch v0.2.28 https://github.com/sst/opencode.git
# already have the repo? check out this version:
$ git checkout v0.2.28

• ›New API endpoint for retrieving GitHub App tokens, enabling programmatic access to GitHub App authentication within OpenCode.

OpenCode v0.2.23 adds modes support, a --mode CLI flag, token/cost session header, and env/file config pointers.

$ git clone --branch v0.2.23 https://github.com/sst/opencode.git
# already have the repo? check out this version:
$ git checkout v0.2.23

$ opencode run --mode research

• ›Adds 'modes' to the TUI, allowing practitioners to switch between configured operational modes during a session.
• ›New --mode flag on the opencode run command lets you specify a mode at invocation time, which is also passed through to the TUI.
• ›Displays token usage and cost information in the session header for real-time spend awareness.
• ›Raises the max output token limit to 32,000, unlocking larger response payloads.
• ›Supports environment variable and file pointers in config, enabling secrets and paths to be sourced externally.

OpenCode v0.2.19 adds @symbol attachments, smarter /editor behavior, and subscription cost hiding in the TUI.

$ git clone --branch v0.2.19 https://github.com/sst/opencode.git
# already have the repo? check out this version:
$ git checkout v0.2.19

$ /editor

• ›Adds @symbol attachments in the TUI for referencing symbols directly in chat input.
• ›Hides cost display when using a subscription model so billing noise doesn't clutter the UI.
• ›Changes /editor auto-send behavior to place content into the input box instead of sending immediately, giving users a chance to review before submitting.
• ›Substitutes the current working directory's home path with ~ in the status bar for cleaner display.

OpenCode v0.2.14 adds configurable log levels, server-side logging, and smarter tsserver spawning.

$ git clone --branch v0.2.14 https://github.com/sst/opencode.git
# already have the repo? check out this version:
$ git checkout v0.2.14

• ›Moves TUI logging to server logs for centralized log visibility.
• ›Adds configurable log levels for tunable verbosity.
• ›Adds Discord redirect integration.
• ›LSP now spawns only a single tsserver per project root, reducing resource overhead.

OpenCode v0.2.9 adds Zig Language Server (ZLS) support and improves config JSON schema with defaults and examples.

$ git clone --branch v0.2.9 https://github.com/sst/opencode.git
# already have the repo? check out this version:
$ git checkout v0.2.9

• ›Adds support for the Zig Language Server (ZLS), enabling Zig-aware code intelligence in OpenCode sessions.
• ›Config JSON schema now declares default values and examples for improved in-IDE documentation and autocomplete.

OpenCode v0.2.6 adds --model/--prompt flags, image/PDF paste, and command aliases to the TUI.

$ git clone --branch v0.2.6 https://github.com/sst/opencode.git
# already have the repo? check out this version:
$ git checkout v0.2.6

$ opencode --model anthropic/claude-opus-4-5 --prompt "Review this code for security issues"

• ›Adds --model and --prompt flags to the TUI for specifying model and initial prompt at launch.
• ›Enables pasting images and PDFs directly into the TUI.
• ›Adds command aliases in the TUI for faster navigation and actions.

OpenCode v0.1.192 adds file attachment support in the TUI, including rendering of PDFs and images.

$ git clone --branch v0.1.192 https://github.com/sst/opencode.git
# already have the repo? check out this version:
$ git checkout v0.1.192

• ›Adds file attachment support in the TUI, enabling users to attach files directly in the chat interface.
• ›Renders attached files (including PDFs and images) inline within the TUI.
• ›Removes the banned command concept, expanding the range of commands available to the AI.

OpenCode v0.1.182 adds an unshare command to the TUI.

$ git clone --branch v0.1.182 https://github.com/sst/opencode.git
# already have the repo? check out this version:
$ git checkout v0.1.182

• ›Adds unshare command in the TUI to revoke sharing of a session or resource.

OpenCode v0.1.177 adds a TUI file viewer and message selection capability.

$ git clone --branch v0.1.177 https://github.com/sst/opencode.git
# already have the repo? check out this version:
$ git checkout v0.1.177

• ›New TUI file viewer lets practitioners browse and inspect files directly within the OpenCode interface.
• ›New message selection capability in the TUI enables selecting individual messages in a session.

OpenHands 0.50 adds Moonshot AI Kimi-K2 model support and new CLI confirmation dialog options

$ git clone --branch 0.50.0 https://github.com/All-Hands-AI/OpenHands.git
# already have the repo? check out this version:
$ git checkout 0.50.0

• ›Adds Moonshot AI Kimi-K2 (kimi-k2-0711-preview) as a recommended and natively supported model via the OpenHands provider.
• ›Adds new options to CLI confirmation dialogs, giving practitioners more control over interactive approval flows.
• ›Persists alias choices in the CLI across sessions, reducing repetitive configuration.
• ›Improves MCP settings UI layout and clarity for easier Model Context Protocol configuration.

OpenHands 0.49 adds CLI/VSCode integration, a new Memory UI, OpenHands Cloud LLM provider, and conversation management improvements.

$ git clone --branch 0.49.0 https://github.com/All-Hands-AI/OpenHands.git
# already have the repo? check out this version:
$ git checkout 0.49.0

$ /settings

• ›Adds CLI/VSCode integration, enabling OpenHands to be used directly from the command line and VS Code.
• ›Introduces an OpenHands Cloud LLM provider, letting users route model calls through OpenHands Cloud.
• ›New Memory UI feature surfaces agent memory state in the interface.
• ›Conversation cards now display branch name and git provider at a glance.
• ›Suggested tasks are now split and organized by git provider.

OpenHands 0.48 adds user-directory microagents, .cursorrules support, conversation stop control, and setup.sh event visibility.

$ git clone --branch 0.48.0 https://github.com/All-Hands-AI/OpenHands.git
# already have the repo? check out this version:
$ git checkout 0.48.0

• ›Supports loading microagents from the user directory (~/.openhands/microagents/) for per-user agent customization without touching project repos.
• ›Supports .cursorrules files placed in the project root directory, letting teams reuse existing Cursor editor rule sets.
• ›Adds ability to stop in-progress conversations from the UI.
• ›Surfaces setup.sh script execution in the event stream so users can observe when and how the setup script runs.

Keymap Editor gains smarter keystroke search with ordered-run matching and repeat-modifier support.

$ git clone --branch v0.196.7 https://github.com/zed-industries/zed.git
# already have the repo? check out this version:
$ git checkout v0.196.7

• ›Keymap Editor now supports searching for repeat modifiers, such as bindings containing cmd-shift cmd.
• ›Keymap Editor keystroke search now matches based on ordered (not necessarily contiguous) runs — e.g., searching cmd-shift-j matches cmd-k cmd-shift-j alt-q and cmd-i g shift-j, but not alt-k shift-j or cmd-k alt-j.

Agent panel now automatically retries failed requests under more circumstances.

$ git clone --branch v0.196.6 https://github.com/zed-industries/zed.git
# already have the repo? check out this version:
$ git checkout v0.196.6

• ›Agent panel automatically retries failed requests under more circumstances.

Zed v0.196.5 adds a keymap editor, debugger memory view, data breakpoints, and new workspace command palette actions.

$ git clone --branch v0.196.5 https://github.com/zed-industries/zed.git
# already have the repo? check out this version:
$ git checkout v0.196.5

{
  "show_menus": true
}

{
  "envFile": "${workspaceFolder}/.env"
}

• ›Adds project panel: rename, project panel: delete, and project panel: duplicate actions to the workspace command palette, enabling file operations when focused on the workspace.
• ›Adds editor: toggle focus action to jump to the last active editor pane item.
• ›Adds severity argument to editor::GoToDiagnostic, editor::GoToPreviousDiagnostic, project_panel::SelectNextDiagnostic, and project_panel::SelectPrevDiagnostic actions for filtering navigation by diagnostic severity.
• ›Adds show_menus setting (Linux/Windows) to always show the menu bar.
• ›Adds panel.sticky_entry.background theme selector for styling sticky project panel entries.

Zed v0.195.4 adds auto-retry in Burn Mode and background-completion notifications for the Agent.

$ git clone --branch v0.195.4 https://github.com/zed-industries/zed.git
# already have the repo? check out this version:
$ git checkout v0.195.4

• ›Auto-retries Agent requests automatically when Burn Mode is enabled.

Zed v0.195.2 adds xAI support, sticky project panel scroll, new agent/debugger settings, and terminal contrast adjustment.

$ git clone --branch v0.195.2 https://github.com/zed-industries/zed.git
# already have the repo? check out this version:
$ git checkout v0.195.2

// In your Zed settings.json:
{
  "terminal": {
    "keep_selection_on_copy": true
  }
}

// In your Zed settings.json:
{
  "agent": {
    "expand_terminal_card": false,
    "expand_edit_card": false
  }
}

// In your Zed settings.json:
{
  "sticky_scroll": false
}

• ›Adds "sticky_scroll": false setting to disable sticky scroll in the project panel, which keeps parent directories visible while scrolling.
• ›Adds agent.expand_terminal_card setting to control whether terminal cards in the agent panel show or hide full command output.
• ›Adds agent.expand_edit_card setting to control whether edit cards in the agent panel show or hide the full diff of a file's changes.
• ›Adds drag_and_drop_selection.delay_ms setting to configure text drag-and-drop behavior in the editor.
• ›Adds keep_selection_on_copy terminal setting (default: false); set to true to preserve text selection after copying.

Zed v0.194.3 adds SVG previews, regex error reporting, dock-size actions, NO_PROXY support, and debugger improvements.

$ git clone --branch v0.194.3 https://github.com/zed-industries/zed.git
# already have the repo? check out this version:
$ git checkout v0.194.3

{
  "adapter": "CodeLLDB",
  "sourceLanguages": ["rust"],
  "program": "./target/debug/my_binary"
}

• ›Adds workspace::DecreaseOpenDockSize, workspace::IncreaseOpenDockSize, workspace::ResetOpenDockSize, workspace::DecreaseActiveDockSize, workspace::IncreaseActiveDockSize, and workspace::ResetActiveDockSize actions for programmatic dock size control.
• ›Adds GEMINI.md as a supported agent rules file name alongside existing rules files.
• ›Adds attachSimplePort to the JavaScript DAP schema for JavaScript debugger configurations.
• ›Respects the NO_PROXY environment variable when any HTTP proxy is configured.
• ›Adds SVG file preview accessible via the quick action bar or keyboard shortcuts (ctrl/cmd+k v and ctrl/cmd+shift+v) when editing SVG files.

Zed v0.193.3 adds Helix mode, Vercel AI provider, debugger variable watchers, and MCP server management improvements.

$ git clone --branch v0.193.3 https://github.com/zed-industries/zed.git
# already have the repo? check out this version:
$ git checkout v0.193.3

{
  "helix_mode": true
}

{
  "dap": {
    "args": ["--log-level", "debug"]
  }
}

{
  "collapse_untracked_diff": true
}

• ›Adds helix_mode setting to enable/disable Helix key bindings (enabling helix_mode also enables vim_mode).
• ›Adds collapse_untracked_diff setting to auto-collapse untracked diffs in the Git panel.
• ›Implements dap.args setting to pass custom arguments to a debug adapter binary.
• ›Adds editor::ConvertIndentationToSpaces and editor::ConvertIndentationToTabs actions to change editor indentation style.
• ›Adds optional clone: bool parameter (default: false) to workspace::MoveItemToPane and workspace::MoveItemToPaneInDirection to clone items into destination panes instead of moving them.

gemini-cli v0.1.15 adds IDE integration, MCP OAuth, loop detection, proxy support, custom themes, and more.

$ git clone --branch v0.1.15 https://github.com/google-gemini/gemini-cli.git
# already have the repo? check out this version:
$ git checkout v0.1.15

hideBanner: true

$ /ide install
/ide status

• ›Adds hideBanner setting to disable the startup banner.
• ›Introduces a VSCode companion extension for IDE integration, with /ide status and /ide install commands to manage it.
• ›Adds Zed editor integration.
• ›Introduces a loop detection service that identifies and breaks simple agentic loops, including an LLM-based loop check.
• ›Adds MCP OAuth infrastructure (Part 1) to support authenticated MCP servers.

gemini-cli v0.1.13-nightly adds MCP server filtering flags, SVG support, extension listing, Cloud Shell auth reuse, and per-server tool include/exclude controls.

$ git clone --branch v0.1.13-nightly.250730.091804c7 https://github.com/google-gemini/gemini-cli.git
# already have the repo? check out this version:
$ git checkout v0.1.13-nightly.250730.091804c7

$ gemini --allowed-mcp-server-names my-safe-server,audit-server

# In settings.json or .gemini/settings.json
{
  "mcpServers": {
    "my-server": {
      "command": "npx my-mcp-server",
      "includeTools": ["read_file", "list_dir"]
    }
  }
}

$ gemini --list-extensions

• ›Adds --allowed-mcp-server-names flag to restrict which MCP servers are loaded at startup.
• ›Adds excludeTools and includeTools per-server config options in mcpServers to control which MCP tools are exposed.
• ›Adds a command-line option to enable and list extensions.
• ›Adds .svg file support for inline content handling.
• ›Enables Gemini CLI to reuse the user's existing auth in Google Cloud Shell.

gemini-cli v0.1.13-nightly adds MCP tool filtering, extension listing, SVG support, Cloud Shell auth reuse, and startup warnings.

$ git clone --branch v0.1.13-nightly.250729.83c4dddb https://github.com/google-gemini/gemini-cli.git
# already have the repo? check out this version:
$ git checkout v0.1.13-nightly.250729.83c4dddb

$ gemini --allowed-mcp-server-names filesystem,github

{
  "mcpServers": {
    "filesystem": {
      "command": "npx",
      "args": ["-y", "@modelcontextprotocol/server-filesystem", "/workspace"],
      "includeTools": ["read_file", "list_directory"]
    }
  }
}

$ gemini --list-extensions

• ›Adds --allowed-mcp-server-names flag to restrict which MCP servers are activated at startup.
• ›Adds excludeTools and includeTools options in mcpServers config to filter individual MCP tools per server.
• ›Adds a command-line option to enable and list extensions.
• ›Adds SVG file support for @file content inclusion.
• ›Enables Gemini CLI to reuse the user's existing auth in Google Cloud Shell.