Chaos Intelligence released Chaos, a source-available functional programming language that treats AI agents as callable values within programs. Computation and flow are written as functions while understanding and generation are delegated to agents, letting both pass data and collaborate in one program. It targets developers building goal-driven, observable agentic systems.
A developer created Ingot, an open-source programming language and toolchain for building AI agents, aiming to make agent development feel like writing a small program rather than stitching together prompts, configs, and framework code. Ingot lets developers describe agent flow, tools, permissions, budgets, memory, model requirements, and human interaction directly in the program.
A developer created Ingot, an open-source programming language and toolchain for building AI agents, aiming to make agent development feel like writing a small program rather than stitching together prompts, configs, and framework code. Ingot lets developers describe agent flow, tools, permissions, budgets, memory, model requirements, and human interaction directly in the program.
The Chelis team open sourced Chelis, a statically typed, compiled functional language designed for agent authorship with human supervision. It integrates SMT solvers with the type system so agents can write and prove programs, targeting numerical domains like quant finance. It is built in Rust with experimental GPU support.
The Chelis team open sourced Chelis, a statically typed, compiled functional language designed for agent authorship with human supervision. It integrates SMT solvers with the type system so agents can write and prove programs, targeting numerical domains like quant finance. It is built in Rust with experimental GPU support.
A visual programming layer for composing complex programs like AI agents without directly editing code. LLM agents are built via prebuilt components, and runtime events light up on the canvas, making debugging observable without parsing logs. It targets developers who find code-only agent development hard to compose and debug.
A visual programming layer for composing complex programs like AI agents without directly editing code. LLM agents are built via prebuilt components, and runtime events light up on the canvas, making debugging observable without parsing logs. It targets developers who find code-only agent development hard to compose and debug.
A solo developer announced Shape, an AI software-development agent rebuilt over two months from an agentic IDE. It runs on multiple AI providers, edits files, runs commands, reviews its own changes with a second model, and connects to tools like GitHub, Figma, and Linear. It is live at useshape.org with source on GitHub, and the creator is seeking early feedback.
A solo developer announced Shape, an AI software-development agent rebuilt over two months from an agentic IDE. It runs on multiple AI providers, edits files, runs commands, reviews its own changes with a second model, and connects to tools like GitHub, Figma, and Linear. It is live at useshape.org with source on GitHub, and the creator is seeking early feedback.
A Reddit post shares Dragon-IDE, an open-source alternative to Cursor hosted on GitHub. The editor is built on CodeOSS and uses OpenCode as its agentic harness. Its distinguishing feature is agent teams that communicate with each other inside the editor.
A Reddit post shares Dragon-IDE, an open-source alternative to Cursor hosted on GitHub. The editor is built on CodeOSS and uses OpenCode as its agentic harness. Its distinguishing feature is agent teams that communicate with each other inside the editor.