Back to directory

IonClaw

A cross-platform AI agent orchestrator written in C++ that runs the same engine as a native build on servers, phones and tablets, with a web control panel, scheduler and skills you extend using plain Markdown files.

IonClaw

IonClaw is an AI agent orchestrator built from the ground up in C++, compiled natively for Linux, macOS, Windows, iOS, Android, tvOS and watchOS with no runtime, interpreter or container required. On a server it starts with one command and serves a full web control panel; on a phone or tablet it embeds the same engine and runs entirely on the device, which the project presents as its main distinction from server-only orchestrators.

It runs multiple agents with independent models, tools and workspaces, tracks every task on a live board with full history, and lets a user configure agents, providers, credentials and skills from the browser without writing code. Skills extend an agent using plain Markdown files, agents can browse the web to navigate, click, type and extract data, a scheduler runs cron, interval and one-shot tasks, agents can spawn child agents for parallel work, and an MCP server and client expose or consume other agents' tools.

Provider support includes Anthropic, OpenAI, Gemini, Grok, OpenRouter, DeepSeek, Kimi, Ollama and any OpenAI-compatible endpoint.

More in Agent frameworks

See all 902
  • Recurse

    recurse.run

    A serverless harness that lets a coding agent design, test, and deploy its own specialist sub-agents as tools, MCP servers, or bots, defined by a schema-validated manifest.

  • ProofForge

    shulgin.is-a.dev

    An AI-agent pipeline that decomposes a mathematics problem, proves each piece, and formalizes the result in Lean 4 and Mathlib until the Lean kernel accepts it, with six of its proofs already merged into Google DeepMind's formal-conjectures repository.

  • local-coder

    github.com

    Builds a local coding-agent team on top of Ollama and OpenCode by detecting the machine's hardware, recommending models for seven specialist roles, and testing whether each assigned model can actually use tools, edit a repository and pass verification.