Skip to content

Peli

29 posts by Peli

GPT-Image-1

Three side-by-side square frames, each showing a uniquely posed 8-bit style pixel cat. Each frame visually represents image generation from different AI models, using five flat corporate colors and minimalist geometric backgrounds. The cats are simple, highly pixelated, and visually distinct from one another, with no text or people present, creating a clean, corporate, and comparative visual suitable for a blog.

Our team just launched support for the new OpenAI gpt-image-1 image generation model, now available through both OpenAI’s API and Azure AI Foundry. We compared gpt-image-1 to DALL·E 2 and DALL·E 3 by generating 8-bit pixel cat images using the same prompt. Each model produces distinct visual results, and gpt-image-1 brings its own style and interpretation. This update helps you evaluate how current generative models handle familiar creative tasks while leveraging advances in image synthesis. Try running the same workflows you use for existing models to see how output and prompt handling differ with gpt-image-1.

GitHub Models in GitHub Actions

A clean 8-bit illustration uses five corporate colors to show a pixelated digital key bridging a GitHub logo and an AI robot icon. A blocky arrow links these icons, symbolizing smooth integration and automated workflow. There is no text or depiction of people.

Streamlining your CI/CD processes just got easier: GitHub Actions now lets you use GITHUB_TOKEN for authentication with GitHub Models. By connecting your workflows directly to these AI capabilities, you can sidestep the hassle of managing separate personal access tokens and keep your automations secure and maintainable. This update supports a seamless integration path for teams looking to enhance workflow intelligence within the GitHub ecosystem.

Blog Narration

A clean, corporate workspace scene showing a stylized open laptop on a desk. From the laptop’s screen, a colorful geometric blog post splits into two paths: one morphs into a text document symbol and the other into a bold, wavy audio waveform. Clear, simple arrows and icons show the transformation, with shapes hinting at narration, AI, and summarization, all rendered in 8-bit style with a striking five-color palette. No people, text, or clutter are present.

Transform your blog workflow with AI-driven narration and summarization. This solution takes your markdown files, creates a concise summary, and generates an audio narration using advanced text-to-speech models. The script reads blog content, prompts an AI to extract key points, and designs a tailored voice profile for text-to-speech. It incorporates checks to prevent overwriting files and uses sample voice personalities to guide the narration. By integrating functions like runPrompt and speak, you can turn written posts into engaging audio content, broadening accessibility and reach for your audience.

MCP Intent Validation

An 8-bit style illustration shows a weather tool trying to access a computer file, symbolized by a small cloud and a folder graphic. Two pathways extend from the tool: one pathway shows regular sunny weather updates with a sun icon, while the other displays an alert warning triangle for unauthorized file access. The simple geometric shapes and corporate color palette give the image a clean, untextured digital appearance.

Exploring the integration of intent validation within tool response frameworks can greatly enhance their reliability. By leveraging LLM-as-a-Judge, it’s possible to evaluate whether a tool functions according to its predefined behavior, especially when dealing with potential tool misuse or error scenarios. As demonstrated with a simple weather tool scenario, intent validation can prevent actions that deviate from expected outputs, ensuring greater accuracy and data integrity. This approach not only curtails inappropriate tool interactions but also reinforces the importance of maintaining clear tool descriptions and configurations to mitigate potential risks.

MCP Tool Validation

An 8-bit style corporate-tech illustration featuring a glowing digital lock icon symbolizing security through "tools signature hash" on one side, and a scanner emitting wave-like patterns to represent a "content safety scanner" on the other. Abstract geometric shapes symbolize interconnected servers and tools, all depicted in a muted 5-color palette, creating a clean, structured visual. No people or text are present in the image.

GenAIScript has introduced updates to enhance the security of Model Context Protocol (MCP) tools, addressing vulnerabilities like rug pull attacks, tool poisoning, and prompt injection. With options for tools signature hashing and prompt injection detection via content scanners, these features provide reinforced safeguards to maintain integrity across tool definitions and outputs. Ensure your configurations are up to date for comprehensive protection.

Prompting is the New Scripting: Meet GenAIScript - Yohan Lasorsa - dotJS 2025

A simplified 8-bit illustration of a presentation stage with a coding-inspired backdrop. The design incorporates geometric patterns symbolizing AI and interconnected circuits, presented in a corporate and futuristic style. It uses five distinct colors with bold, clean lines, focusing on tech innovation through abstraction, without characters or text for added minimalism.

Yohan Lasorsa presented GenAIScript at dotJS 2025, showcasing how it streamlines interactions with AI models via scripting. As a seasoned Principal Developer Advocate at Microsoft and open-source contributor, he brings over 15 years of expertise spanning applied research, IoT, web development, and cloud architecture. This talk dives deep into shaping the future of AI-powered solutions.

AST Grep and Transform

The image is an abstract 8-bit style illustration using five colors. It shows a stylized tree with nodes connected in a branching pattern, representing an Abstract Syntax Tree (AST) used in code transformation. Geometric shapes and lines indicate the ideas of parsing and modifying source code. The design is simple and corporate, with no characters or text present.

Automating code updates at scale can be tricky, especially when it comes to maintaining accuracy. Abstract Syntax Trees (AST) offer a powerful solution by allowing you to directly manipulate code structures without worrying about formatting inconsistencies. With tools like ast-grep and LLMs, you can locate, transform, and update code efficiently. This approach is ideal for tasks such as generating or updating function documentation in TypeScript projects. Curious how this works? Explore how AST-driven strategies can streamline your workflow.

MCP Resources

A colorful 2D illustration in an 8-bit retro style, depicting a stylized server setup with abstract data nodes linked by lines. The design uses five corporate colors, forming a geometric pattern that symbolizes data and resource exchange. The interconnected nodes illustrate a protocol for interaction, creating an iconic and corporate atmosphere without any people or text.

The Model Context Protocol (MCP) introduces a powerful way for scripts to provide context to large language models (LLMs) by exposing data through MCP resources. These resources enable servers to share structured content that clients can access and utilize efficiently. By integrating publishResource into your workflows, you can streamline how your applications interact with MCP clients, enhancing context discovery and resolution. Dive deeper into how MCP tools and resources can elevate your systems today.

Blog Images

A retro-style 8-bit illustration features a computer screen with a visible code script. Surrounding it are symbolic representations of image generation: a markdown document, an image prompt morphing into abstract art, and a fading alt-text bubble. The design uses corporate colors, simple geometric shapes, and is limited to five colors, without including people or text.

Automatically generating blog post cover images may sound mundane, but the script behind the process showcases some interesting automation techniques. By transforming the markdown content into an image prompt, generating visuals and alt text, and seamlessly updating metadata, this multi-step workflow demonstrates how small-scale automation can scale efficiently. The result? Abstract, content-driven images that hint at optimization potential. Thought-provoking for content creators considering automation workflows.

Scripts as MCP tools!

Imagine an 8-bit style scene showing a futuristic, tech-heavy landscape. In the center, there's a large, geometric server block labeled "GenAIScript MCP Server," connected to two smaller devices tagged as "script A MCP Tool" and "script B MCP Tool." A line connects "GitHub Copilot Chat" to the main server, representing integration. The design is minimalist with only five colors, emitting a corporate, high-tech feeling without any human figures or text.

The Model Context Protocol (MCP) is reshaping how we approach integration with AI-driven tools. Platforms like GitHub Copilot Chat and Copilot Studio are leading adoption efforts, and GenAIScript is now enabling you to expose scripts as MCP tools, streamlining workflows with smarter decision-making by LLMs. Ready to elevate your development process? Explore the details in the documentation.

Azure AI Search

The image shows a cartoon-like computer interface featuring a "retrieval index" with a magnifying glass icon and several files, which are illustrated with animal icons of cats and dogs. This represents the process of indexing in a vector database. There is a visual flow from the files to the database, using icons to depict cats and dogs for the purpose of similarity search. The style is characterized by bold geometric shapes, using a limited corporate color palette of five colors, and maintaining a simple 8-bit aesthetic.

Discover how GenAIScript simplifies interaction with Azure AI Search for building vector-based search systems. With built-in support for chunking, vectorization, and indexing, you can efficiently implement Retrieval Augmented Generation (RAG) workflows, leveraging embeddings for similarity searches across your data. Learn how to streamline file indexing and query execution using straightforward functions and enhance your AI toolset.

Super Charge Copilot Chat

The image depicts a stylized, 8-bit graphic of a computer screen showcasing GitHub Copilot Chat. Abstract visual elements represent API documentation using small code snippets and blocks, illustrating their integration within the chat interface. The scene conveys a sense of technology and software development, designed with corporate aesthetics using a simple five-color palette, without any people or text.

Here’s a game-changer for boosting GitHub Copilot Chat’s performance with GenAIScript: integrate your entire documentation into the chat session. By compressing the GenAIScript docs to fit into the context window, you enable Copilot Chat to deliver sharper, more context-aware results. This approach takes advantage of the latest support for reusable prompts and local workspace indexing, seamlessly embedding documentation to refine interactions. Curious about setting it up? Dive into the linked guide and unlock its potential.

Make it better!

A retro 8-bit computer screen displays a vibrant geometric interface. Abstract icons symbolize code snippets, and a glowing button labeled "make it better" suggests an enhancement process. The backdrop features a simple, five-color geometric pattern, evoking a futuristic corporate environment.

Harnessing the power of the makeItBetter function in GenAIScript simplifies code refinement by automating improvement loops. By analyzing and enhancing your code in just a few steps, this tool maximizes efficiency without the need for manual optimizations. If you’re diving into AI-driven coding workflows, this approach offers a streamlined way to iterate and elevate your results.

Revamping the views...

The image illustrates a split screen in an 8-bit style, contrasting two environments for the GenAIScript interface. On the left, a computer monitor shows the interface with interactive elements like mermaid diagrams and math notations. On the right, a standalone web browser presents the same interface. The composition is framed with a simple, geometric design using five accent colors, creating a vibe of technological progress in a corporate context.

We’ve enhanced the GenAIScript experience by moving beyond Visual Studio Code’s default Markdown preview. Our new custom webview not only supports additional formats like mermaid diagrams and math but also allows you to access and run scripts directly in your browser or other webview-compatible apps. The transition is seamless, with options to test this alongside the existing features. Explore these updates for a more interactive and flexible workflow.

Let there be videos!

A colorful, simplified illustration shows a futuristic desk setup. There's a large computer screen with visual representations of video and audio, like frames and waveforms. Surrounding the screen are stylized geometric symbols for video and audio tools. The design uses abstract shapes in the background to convey an atmosphere of creativity and innovation, drawing inspiration from 8-bit style. It employs just five bold colors to keep the look clear and corporate, with no people or language visible.

Analyze videos with ease. Traditional LLMs require timestamp-based screenshots, often leading to context window issues. GenAIScript simplifies video and audio analysis by enabling automated frame extraction and transcript integration. Learn how tools and agents streamline these tasks for your scripts in the documentation.

Playground, o1 and DeepSeek

An 8-bit style image shows a sleek, futuristic web application interface on a computer screen. It features icons for AI models like OpenAI o1 and DeepSeek V3 against a user-friendly backdrop. The corporate 5-color palette incorporates geometric shapes to create a digital playground theme, focusing on simplicity and iconic design without people or text.

The latest GenAIScript update introduces a self-hosted Playground app, making it easier to experiment with scripts via a user-friendly interface. It bridges the CLI and Visual Studio Code tools seamlessly. Additionally, support has expanded to include OpenAI’s o1 models and DeepSeek V3, empowering users with versatile integrations for advanced scripting.

Node.JS API

An 8-bit style illustration showcases a stylized Node.JS logo alongside a geometric TypeScript symbol. Abstract gear and circuit patterns connect the two, representing the integration of GenAIScript in a corporate environment. The image uses five colors that highlight a digital, clean, and tech-focused aesthetic, remaining iconic and free of human elements or text.

Introducing a new Node.js API for GenAIScript: now you can integrate GenAIScript’s functionality directly into your TypeScript workflows (v1.83+). Mimicking the CLI’s behavior, the run API simplifies tasks like file summarization while maintaining a lightweight and side-effect-free design. Ready to streamline your automation? Check out the documentation for details.

Fallback Tools

An image in retro 8-bit style showing a digital workspace. On the left, there are icons with lines connecting them, symbolizing the idea of "fallback tools." On the right, geometrical shapes represent binary code processing related to tool-enabled LLMs. The background consists of basic geometric shapes that give a corporate and tech-focused feel, using a consistent 5-color scheme without any human figures or text.

Enhance your LLM’s capabilities with fallback tools. When your chosen model lacks native tool support, fallback tools step in, teaching the model how to utilize external resources for better problem-solving. Automatically enabled for recognized models, this feature ensures efficient workflows and can be manually configured for broader compatibility. Stay ahead by integrating smarter solutions.

LLM Agents

A retro, 8-bit style refrigerator opens to reveal unusual items: a toaster, a television, and a penguin. The geometric design and corporate color palette of five colors create a minimalist and iconic scene, devoid of people or text.

GenAIScript redefines how agents interact with users by integrating inline prompts and tools to enhance task execution. In this guide, we explore building a user-interaction agent that actively seeks user input, confirms decisions, and adapts based on user responses. By defining clear metadata, flexible agent behavior, and model configurations, developers can create agents that are both dynamic and intuitive. This approach not only simplifies user-agent communication but also emphasizes context-driven interactions for more accurate outputs. Check the GitHub link included for real-world implementation details.

Listen to the podcast

A digital artwork featuring an 8-bit style geometric design symbolizing digital innovation and technology with a corporate flair. Key elements include a stylized microphone, pixelated soundwaves, and a retro-themed computer, all presented in a simplified color palette of five hues. The focus is on audio and podcast components, merging modern and classic tech aesthetics without any text or human figures.

Learn how we used Google’s NotebookLM to automate podcast creation, simplifying the process of transforming written content into audio. Listen to the result and explore how innovative tools can streamline your workflows.

Keeping your README Fresh and Engaging

A simplistic, pixelated image shows a computer screen displaying a brightly colored README.md file. Surrounding the computer are digital icons depicting various folders and files, symbolizing documentation and examples. The setting is a stylized corporate office, depicted with minimal geometric shapes and a 5-color palette, giving it a classic, retro aesthetic.

A strong README can define your project’s success in the open source world. The GenAIScript used in the GenAI project is a great example of automating README maintenance by pulling the latest details from documentation and samples. This ensures your project’s first impression stays both current and engaging. Curious about how to run it? The GenAIScript CLI makes it seamless. Keep your contributions visible, your documentation sharp, and your users informed.

Unlocking the Power of Prompts - A Gentle Introduction to GenAIScript 🚀

Screenshot of GenAIScript in the VS Code marketplace

Curious about integrating Large Language Models into your workflows? Discover how GenAIScript streamlines prompt creation, automates structured results, and supports various LLMs like OpenAI and GitHub Models. From adding context to leveraging metadata, it provides flexibility to fit your unique needs. Learn more about optimizing your interactions with AI tools effectively.