rlm

IronClaw is an Agent OS focused on privacy, security and extensibility

A feature rich implementation of Recursive Language Models, with ACP integration, REPL tool support, structured IO, advanced visualization, logging tools.

Production focused Self-harnessed LM runtime (RLM) that allows the LM to call its sub-lm with DSPy signatures. Define your inputs, outputs, and tools — the model handles its own control flow. Get fully interpretable trajectories and performance that scales directly with model improvements. Without context rot.

A recursive coding agent inpired by RLMs

Automated generation of comprehensive Agents.md for LLMs, driven by the DSPy Recursive language model implementation.

Autolith is a self-modifiable general purpose Lisp AI agent

Skill + MCP server to turn your agent into an RLM. Load context, iterate with search/code/think tools, converge on answers.

A programmable tool and agent runtime for Pi

中文UI / A terminal-based coding agent with unlimited concurrency, multi-agent orchestration, and a persistent Python runtime (RLM) for stateful data work and recursive subagent delegation

the extensible, customizable, self-documenting, real-time multi-agent computing environment

Source-available autonomous AI agent with a self-model that actually changes as it runs - ego, affect and identity grounded in real psychology (Higgins, PAD/OCC). Writes its own tools, drives a real browser, runs companies, and grades itself on measured outcomes. Stops before anything it can't take back.

Official Implementation of "Reasoning Language Models: A Blueprint"

CLI for Recursive Language Models

Replace LLMs reasoning with code execution to solve real-world tasks.

Traffic light for AI Agents and TypeScript/Node multi-agent orchestrator with shared state, guardrails, and adapters for 32 AI frameworks

Rust CLI implementing the Recursive Language Model (RLM) pattern for Claude Code. Process documents 100x larger than context windows through intelligent chunking, SQLite persistence, and recursive sub-LLM orchestration.

Agentic RLM workbench on DSPy 3.3: sandboxed code execution, live operator streaming over SSE, durable sessions & artifacts, terminal-first client

The Research Playground for the RLMSs and Coding Agents

A drop-in code interpreter using the Monty Python emulator for DSPy's RLM module.

Claude Code skill for handling long-context tasks through recursive decomposition