BuildRiot
Sign in

AISource: AWS

AWS adds four managed Airflow agent skills; upgrade skill supports Provisioned only

AWS added four agent skills for coding assistants to author, debug and test managed Airflow workflows and upgrade environments. Authoring, debugging and testing support Provisioned and Serverless; the upgrade skill supports Provisioned only.

By News DeskPublished here (UTC)2 min read
AWS logo
Cover: BuildRiot

AWS announced four Amazon Managed Workflows for Apache Airflow agent skills through Agent Toolkit for AWS on October 15, 2026. AWS says the skills are available across all AWS Regions offering the service. Its documentation lists Kiro, Claude Code, Amazon Q, Cursor and Codex as compatible coding assistants. Sources: AWS

The authoring-mwaa-workflow skill authors and deploys Python directed acyclic graphs (DAGs) for Provisioned environments. It also authors and deploys YAML workflows for Serverless. Both environments support debugging-mwaa-workflow for diagnosing failures. They also support testing-mwaa-workflow for triggering deployed workflows, monitoring completion and retesting fixes. Although the announcement broadly describes support for both environment types, the MWAA user guide specifies that upgrading-mwaa-environments supports Provisioned only. It offers blue-green or in-place upgrades, including Apache Airflow 2.x to 3.x. Sources: AWS

AWS says the skills reduce manual work and generate working DAGs from natural-language descriptions. The skills execute AWS CLI commands. AWS recommends its Model Context Protocol (MCP) Server. It provides current documentation, sandboxed execution and audit logging. Sources: AWS

AWS documents resource-scoped Identity and Access Management (IAM) permissions. These include airflow for environments and the Airflow REST API, and airflow-serverless for Serverless workflows and runs. MCP forwards the caller’s credentials for authorization by the destination service. Session policies and permission boundaries also constrain temporary credentials. Separate MCP invocation grants are unnecessary. The aws-mcp:InvokeMcp, aws-mcp:CallReadOnlyTool and aws-mcp:CallReadWriteTool actions have no effect, including in Deny statements. Sources: AWS

AWS directs users relying on legacy Deny statements to replace them with aws:ViaAWSMCPService or aws:CalledViaAWSMCP conditions. These conditions belong in IAM policies or service control policies. The Boolean aws:ViaAWSMCPService condition indicates that a request passed through an AWS managed MCP server. The aws:CalledViaAWSMCP condition identifies the server’s service principal. For AWS MCP Server, that principal is aws-mcp.amazonaws.com. Sources: AWS

What to do

  1. If you rely on Deny statements using these legacy aws-mcp actions, replace them with downstream service controls using aws:ViaAWSMCPService or aws:CalledViaAWSMCP.
  2. Scope airflow and airflow-serverless permissions to the environments and workflows the assistant needs.
  3. For the Claude Code plugin and MCP setup, install uv for the proxy and configure local AWS IAM credentials before executing API calls or scripts.
  4. In Claude Code, run /plugin install aws-data-analytics@claude-plugins-official, then /reload-plugins.
  5. Check the MCP connection by starting a new agent conversation and asking which AWS Regions are available. Sources: AWS

Why it matters

Airflow teams gain an option to operate workflows through compatible coding assistants. The environment-upgrade skill supports Provisioned only, and MCP access depends on downstream service permissions rather than legacy invocation restrictions.

Attribution

Sources

  1. 1AWSAmazon Managed Workflows for Apache Airflow now offers AI agent skillsSource published (UTC)
  2. 2AWSUsing agent skills with Amazon MWAA
  3. 3AWSHow AWS MCP Server works with IAM
  4. 4AWSGetting started - Agent Toolkit for AWS

Read the original source for full details and context.

Report a problem

Links, dates or what you expected. Do not include passwords or private data.

More on AI

All AI stories

Review · AI

Elastic’s Linux profiling preview: Setup, requirements and limits

Elastic says its OpenTelemetry Profiling integration collects continuous CPU profiles on Linux through eBPF. It requires no application instrumentation or restarts, according to Elastic. Version 0.0.1 is a technical preview. OpenTelemetry Profiles remains Alpha. Elastic’s October 14, 2026, walkthrough performs the one-time Universal Profiling storage initialization as superuser. It then configures a Fleet policy with samples_per_second=20 and enrolls the Linux agent. To verify collection, Elastic directs users to Infrastructure > Universal Profiling > Flamegraphs. Users select the traffic window and filter process.executable.name: "java"; agent health alone is insufficient. Elastic reports 30,150 PetClinic requests in four minutes at slightly over half a CPU core. The walkthrough does not report a profiling-overhead measurement.

Elastic

Review · AI

Nemotron Bolt claims a 24GB footprint; its deployment recipes target larger GPUs

NVIDIA positions Nemotron Bolt for local agentic work, including coding, with a stated 24GB memory footprint. Its published GGUF deployment recipes target discrete GPUs with at least 32GB or 48GB of memory, depending on configuration. NVIDIA announced availability on October 15, 2026. The quantized model card lists 74 billion total parameters and 9 billion active parameters. It supports multimodal inputs and text output. It specifies mixed quantization: IQ2-XXS experts, NVFP4 Mamba layers and Q8_0 embeddings and output. The setup instructions use llama.cpp with a separate LiLiCorr drafter for speculative decoding.

NVIDIA

AWS adds four managed Airflow agent skills; upgrade skill supports Provisioned only · BuildRiot