Recent blog posts reveal strong trends toward workflow automation and the use of multi-agent architectures powered by advanced SDKs and orchestration tools. Key announcements include the introduction of dynamic and automated workflows (e.g., Jira–GitHub integration via codex-cli, Claude dynamic workflows), which streamline processes, reduce manual effort, and facilitate complex, parallelized tasks. Both OpenAI and Anthropic are promoting agent-based solutions that leverage their latest APIs (Deep Research API, Claude Agent SDK), supporting multi-agent orchestration, integration with external systems, and advanced automation in research, development, and operational scenarios. These developments emphasize efficiency gains, real-time information processing, and easy extensibility across coding, research, and reliability engineering workflows.
New Cookbook Recipes
jira-github.ipynb
Source: openai/openai-cookbook
The blog post introduces a cookbook designed to automate the workflow between Jira and GitHub using the codex-cli tool integrated within a GitHub Action. This automation starts when a Jira issue is labeled, triggering a process that creates a GitHub pull request (PR), updates both platforms, and facilitates code review with minimal manual effort. Key components include:
- Jira Automation Rule: Activates on label changes and communicates with GitHub to initiate workflows.
- GitHub Action Setup: Runs a predefined YAML workflow; it processes the Jira issue, uses
codex-clito implement ticket requirements, and creates a PR. - End-to-End Flow: Upon labeling, the Jira issue transitions statuses automatically, and relevant comments are added to the ticket.
The integration enhances efficiency, reduces manual updates, and streamlines development processes.
introduction_to_deep_research_api_agents.ipynb
Source: openai/openai-cookbook
The “Deep Research Agents Cookbook” introduces users to building advanced research workflows utilizing the OpenAI Deep Research API and Agents SDK. Key features include the orchestration of single and multi-agent pipelines, enrichment of user queries for optimized output, integration of web search and internal file search, and streaming of research progress. Users are instructed to set up their environment with prerequisites such as the OpenAI API key and relevant SDKs.
A noteworthy addition is the “Basic Research Agent,” which employs the gpt-5.6-terra model for deep empirical research, streaming real-time results. A multi-agent architecture is also presented to enhance research quality through a triage process involving clarifying and instructional agents. The cookbook emphasizes that Deep Research is suited for complex tasks requiring planning and reasoning rather than simple inquiries.
08_Dynamic_workflows.ipynb
Source: anthropics/claude-cookbooks
The blog post introduces “dynamic workflows,” a new feature in Claude Code designed for managing complex tasks requiring multiple agents. This feature allows users to write JavaScript orchestration scripts that enable agents to operate in parallel or in stages, effectively handling tasks that exceed the limits of a single context window. Dynamic workflows enhance the efficiency of fact-checking activities, such as verifying investor updates against source documents.
Key features include the capability to spawn up to 1,000 agents while running up to 16 concurrently, providing flexibility in selecting models for different stages of a task. Workflows can be triggered via the Python Agent SDK, ensuring easy integration. This functionality is particularly beneficial for scenarios requiring structured verification and detailed orchestration. Dynamic workflows are available on all paid plans with access to the Anthropic API and across major cloud platforms.
README.md
Source: anthropics/claude-cookbooks
The blog post introduces the Claude Agent SDK, a framework for building advanced general-purpose agents, spanning a tutorial series from simple implementations to complex orchestration systems. It outlines the setup process, which includes installing necessary tools, cloning the project, and obtaining API keys.
Key topics covered in the series include:
- Fundamentals of the SDK, including core interfaces and tool usage.
- Multi-agent orchestration and integration with external systems via the Model Context Protocol (MCP).
- Development of specific agents, such as a Chief of Staff agent, an observability agent for DevOps workflows, and a site reliability engineer agent.
The SDK, initially designed for coding tasks, now serves broader applications like research, data analysis, content generation, and workflow automation, showcasing its versatility in creating efficient agents across various domains.