Releases·Agent Release
Aex opensources Brain, a Rust agent runtime with Wasm agent loops and stateful sessions
The Brain team open-sourced Brain, an MIT-licensed Rust runtime for persistent agent services, in an early preview announced on Reddit seeking feedback. Brain separates the agent decision loop, which compiles to Wasm, from the runtime that handles session state, model and tool I/O, streaming, cancellation, restart recovery, and journaling. It targets agents that have outgrown simple scripts, where infrastructure rather than the model loop is the hard part.
6 engagement2 sourcesWed, Sep 2, 2026, 10:44 AM
AI Overview
- One Rust binary owns the session while the agent's decision loop runs in Wasm, with the runtime performing all model and tool I/O
- Typed tools can run in the app process, a browser, a microVM, or a remote backend without changing the session API
- Includes a TypeScript SDK and support for 70+ model providers; repo at github.com/aexhq/brain under MIT license
- Currently an early preview; the author disclosed working on Brain and asked for feedback on whether the 'runtime for agents that have outgrown scripts' positioning is clear
- Motivation: once agents run as persistent services, session state, restarts, streaming, cancellation, and tool execution become the actual engineering work
Entities
Related
Releases·Tool Release
Show HN: KISS – A Rust agent harness inspired by Pi, with sub-agents, MCP, and browser WASM builds
A developer launched KISS on GitHub, a performant agent harness written in Rust and modeled on the Pi harness, created after frustration with Pi's performance. It targets practitioners who want an embeddable, multi-language agent runtime.
A developer launched KISS on GitHub, a performant agent harness written in Rust and modeled on the Pi harness, created after frustration with Pi's performance. It targets practitioners who want an embeddable, multi-language agent runtime.
4 engagement1 sourcehackernewsSat, Sep 26, 2026, 12:05 AM
Releases·Tool Release
Dev releases Brain, an open-source minimal agent runtime pitched against LangChain and Mastra (<1ms session, ~25MB binary)
A developer open-sourced 'brain' (github.com/aexhq/brain), a minimal, extensible agent runtime inspired by Anthropic's managed agents engineering post and pi.dev. It targets production agent use cases the author found too rigid in LangChain and Mastra. For practitioners, it offers a lightweight alternative with sub-second cold starts and a small compiled footprint.
A developer open-sourced 'brain' (github.com/aexhq/brain), a minimal, extensible agent runtime inspired by Anthropic's managed agents engineering post and pi.dev. It targets production agent use cases the author found too rigid in LangChain and Mastra. For practitioners, it offers a lightweight alternative with sub-second cold starts and a small compiled footprint.
2 engagement1 sourceredditMon, Sep 7, 2026, 10:08 PM
Products·launched
Rebuno - open-source execution runtime for operating AI agents in production
Rebuno is an open-source runtime that manages execution state and guardrails for AI agents in production, regardless of the framework they were built with. Agents submit tool and model calls as steps before executing them, and Rebuno evaluates each step against per-agent YAML policies, allowing, denying, or holding calls for human approval. It solves the problem of safely operating agents built with different frameworks in production environments.
Rebuno is an open-source runtime that manages execution state and guardrails for AI agents in production, regardless of the framework they were built with. Agents submit tool and model calls as steps before executing them, and Rebuno evaluates each step against per-agent YAML policies, allowing, denying, or holding calls for human approval. It solves the problem of safely operating agents built with different frameworks in production environments.
3 engagement1 sourcehackernewsMon, Sep 14, 2026, 04:29 PM
Products·launched
Open-source AI agent platform built with Bun and TypeScript
Cybara is a fully open-source, MIT-licensed AI agent platform built with TypeScript and Bun. It combines agents, tools, plugins, skills, MCP, ACP, LSP, browser and desktop automation, model routing, provider plan tracking, messaging integrations, memory, multi-agent workflows, and web, terminal, desktop, and mobile interfaces for all OS platforms. It solves the problem of needing a single self-hosted stack to control agents, tools, models, and runtime.
Cybara is a fully open-source, MIT-licensed AI agent platform built with TypeScript and Bun. It combines agents, tools, plugins, skills, MCP, ACP, LSP, browser and desktop automation, model routing, provider plan tracking, messaging integrations, memory, multi-agent workflows, and web, terminal, desktop, and mobile interfaces for all OS platforms. It solves the problem of needing a single self-hosted stack to control agents, tools, models, and runtime.
2 engagement1 sourcehackernewsThu, Jul 16, 2026, 11:04 AM
Products·launched
Cadenya – managed agent runtime using your APIs/MCP servers as tools
Cadenya is a hosted agent runtime where your existing APIs or MCP servers serve as tools in a managed agentic loop. It is model-provider agnostic, accepting OpenRouter or any OpenAI-compatible endpoint for inference. It targets developers building agents who want tool integration, model flexibility, and frontend deployment without self-managing infrastructure.
Cadenya is a hosted agent runtime where your existing APIs or MCP servers serve as tools in a managed agentic loop. It is model-provider agnostic, accepting OpenRouter or any OpenAI-compatible endpoint for inference. It targets developers building agents who want tool integration, model flexibility, and frontend deployment without self-managing infrastructure.
1 engagement1 sourcehackernewsThu, Sep 10, 2026, 10:52 PM