stateful-agents

Stateful agents that are like people, with memory, identity, and the ability to learn and adapt

High-performance Rust accelerators for LangGraph applications. Drop-in components that provide up to 700x speedups for checkpoint operations and 10-50x speedups for state management.

Agent emotional continuity with PAD state, trust, appraisal, and compact emotion logs.

Research benchmark for evidence-grounded OS-agent collaboration, continuous state diagnosis, scoped memory reuse, and stale-state rejection.

ARC-AGI-2 solver: 95.7% public eval at $3.12/task — lowest cost above 95%. Full inference traces included.

Build your agent once. Carry it everywhere. Compiles portable agent packs for Pi, Codex, Claude Code, and Cursor.

The first open evaluation framework for AI continuity. 250 narrative tests, 1835 verification questions, 10 checkpoints. Benchmark for AI memory systems, stateful agents, and long-term context persistence. No LLM in the evaluation loop.

A cognitive runtime that gives LLM agents persistent state, identity, and learning across turns. Memory, beliefs, drives, self-evolution, skills, and affective state — engineered scaffolding outside the model.

Agent state infrastructure: state-trace Python working memory plus @razroo/parallel-mcp durable MCP orchestration.

LangGraph is a powerful framework built on LangChain that enables the creation of stateful, multi-step, and agentic workflows using directed graphs. It simplifies complex LLM orchestration by allowing conditional branching, memory, and tool integrations in a visual and modular way.

Letta integration for the Ejentum Reasoning Harness. 8 Python functions (4 harnesses × dynamic + adaptive) registered via tools.upsert_from_function.

C# SDK for the Letta API -- stateful AI agents with persistent memory, tools, and multi-agent groups

Letta (formerly MemGPT) is a stateful AI agents platform built around long-term memory, tool execution, and multi-agent coordination. The Letta REST API exposes 239 endpoints across 36 public resource categories — agents, memory blocks, archival memory, sources (RAG), custom tools (sandboxed/client-side/MCP), MCP servers, multi-agent groups…

Structured memory and snapshot history system for AI agents (OpenCode / Claude Code)

Platform for stateful agents: AI with advanced memory that can learn and self-improve over time.

Stateful Goal mode skill for Codex, Claude Code, and AI agents with STATE.md memory, gates, and stop rules.

Dev Partner is a Codex skill for long-running feature development in a local git repository connected to GitHub. It keeps feature work stateful, reviewable, and explicitly controlled by the human developer.

Stateful AI study coach on Cloudflare Agents with persistent plan memory, Workers AI inference, tool calling, streaming chat, and scheduled workflow execution.