Oracle shipped a CLI tool and Visual Studio Code extension for its Fusion AI Agent Studio on July 27, 2026, giving product development teams a local, code-first environment for building, extending, and deploying agentic applications. The release brings workspace-based project management, AI-assisted coding through Codex, and direct access to enterprise APIs - an offering built to match the workflows professional developers already use.
While no-code and guided builders remain available, the pro-code experience puts Fusion AI Studio artifacts - workflows, agents, business objects, and tools - into a local project structure. Teams can organize and version-control these assets just like any other software project.
Three paths to a local workspace
Sets up the CLI workspace using a natural-language prompt in Codex. Developers can instruct Codex to clone the Fusion AI Studio repository, install prerequisites, pull required skills and sample applications, and open the project in VS Code - all while following the official installation guide exactly.
A sample shell script handles the directory scaffolding and copies the necessary AI Studio skills and sample assets into the expected locations. Teams that want a repeatable, auditable onboarding process can download and adapt the script before running it. The script does not install the VS Code extension, so developers complete that step manually afterward.
Developers who prefer step-by-step control can follow the Fusion AI Studio GitHub repository and the Configuration guide directly. All three setup paths converge on the same local layout: a workspace folder that holds shared skills, sample AI apps, and project directories like AiCoE-BootCamp.
Product teams that already work with AI Agents & Automation will recognize the value of a repeatable local environment. The local workspace and Codex integration give developers the same kind of edit-test-publish loop they use for traditional software, reducing the distance between ideation and a validated workflow.
Building a 'Hello World' agentic workflow
The CLI's first-run experience mirrors the classic developer convention. After opening the project folder in VS Code and running aistudio init, a scaffold of starter files appears. Authentication is configured by providing the Fusion AI Studio server URL and CLI credentials - information that gets stored in the workspace's env.properties file.
With the environment connected, developers create a new workflow either through VS Code commands or by fetching existing artifacts from a server. In a demonstration walkthrough, Codex uses a plain-English prompt to generate a simple "Hello World" workflow: it greets the user by name and delivers a short motivational message, all under 20 words, with HCM and Global HR set as the business context.
The CLI automatically runs test scripts during validation, and the workflow can be reviewed, refined, and accepted directly in the editor. The extension tracks generated changes, so developers can inspect what Codex proposes before merging it into the project.
Oracle's AI Center of Excellence designed the extension to treat natural-language collaboration as a peer to manual editing. "Keep an eye on those sneaky notifications - they love hiding the authentication pop-up you actually need to click," the team noted in the setup guidance, a reminder that even AI-native tooling still relies on conventional OAuth flows.
Why this matters for product development
The Fusion AI Agent Studio CLI brings agentic application development into the tools and revision control systems that product teams already manage. Instead of building workflows in a separate UI and then exporting them, teams can create, test, and version AI assets alongside the code that calls them. For product development groups that ship frequent updates, this local, scriptable workflow cuts the time between a requirement and a deployable change - and makes the entire process auditable and reviewable.
Your membership also unlocks:
