recursive-language-models

Method for Long Context RLMs using verifiable Lambda Calculus

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

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

The Research Playground for the RLMSs and Coding Agents

Implementation of Recursive Language Model paper from scratch

🫛👣 FAVA Trails — Git-native, curated memory for AI agents via MCP. Draft isolation, promotion gate, thought lifecycle hooks, memory curation protocols, supersession chains.

CLI agent that analyses meeting transcripts using Recursive Language Models (RLMs) to extract decisions, ownership, deadlines, and contradictions across documents.

Running Recursive Language Models (RLM) on Azure

AccInt - local-first MCP Work Model for coding agents that learns from real outcomes.

Machine-checked Lean 4 proofs for "Recursive Language Models Through the Admissibility-Dynamics Framework." Covers RLM sub-call architecture, three sufficient conditions for bounded-inconsistency deployment (safe abstention, bounded-decomposable predicates, runtime depth verification), training class closure, and the deployment-boundary synthesis.

Recursive Language Models — process arbitrarily long prompts by offloading context into a REPL with symbolic recursion via sub-LLM calls. Based on Zhang, Kraska & Khattab (2025).

Zig runtime for Recursive Language Model workflows, inspired by alexzhang13/rlm

Stateful agent on a Recursive Language Model engine — a local LLM drives an IPython REPL, queries sub-models from code, and consolidates a durable memory each episode.

A verifiers RLM environment for testing whether adaptive recursive search outperforms brittle manual RAG choreography on long synthetic corpora.

Recursive Language Model - answers questions by navigating a document tree instead of reading it whole; bounded context per call, OpenAI & Gemini

Chain-of-density study of 'Recursive Language Models' (Zhang, Kraska & Khattab 2025, MIT CSAIL, arXiv:2512.24601) - five-tier note, locator-verified. The prompt-as-environment paradigm Trellis implements: REPL offloading, recursive sub-calls, 10M+ token inputs.

Recursive Language Models — the root LLM gets a Python sandbox and delegates focused sub-questions to cheaper sub-models, then synthesizes. Includes a finance-sector demo on a synthetic 10-K filing.

Portable Recursive Language Models implementation — drive it from any AI coding assistant (Claude Code, Copilot, Cursor, Aider) or import as a Python library. Offloads large inputs to a Python REPL and distributes semantic work across sub-LLM calls. Multi-provider: Anthropic, OpenAI, Azure, OpenRouter, Ollama, LM Studio, GitHub Models.

A local, sandboxed prototype of Recursive Language Models (RLMs) using Docker and Ollama.

Analyze entire SEC filings with Recursive Language Models. Fetch from EDGAR, ask questions that need the whole document, get cited answers. Open source CLI and Python library, MIT licensed.