nvlabs/spatialclaw
SpatialClaw: Rethinking Action Interface for Agentic Spatial Reasoning
01 / MACHINE INTELLIGENCE
Models you can run. Agents you can inspect. Tools you can build on.
← All topicsSpatialClaw: Rethinking Action Interface for Agentic Spatial Reasoning
Ask several AI models one question — they answer in parallel, critique each other blind, and you decide. Native macOS, bring-your-own-keys.
Research artifacts from Recursive's automated AI research system
See exactly what you spend on Claude CLI + cost optimizations functions
Your AI agent in 30 seconds. Not 30 hours. Self-hosted, open-source personal AI with desktop installer, multi-agent Command Center(Deep Work), and 7-layer security. Anthropic, OpenAI, or Ollama.
🚀🚀🚀 [ICML 2026] Official Implementation of Robust-U1: Can MLLMs Self-Recover Corrupted Visual Content for Robust Understanding?
Use agent to learn agent - A skeleton course on how to design, build, and operate production AI agents - GitHub - bryanyzhu/agentic-ai-system-course: Use agent to learn agent - A skeleton course on how to design, build, and operate production AI agents
Cognitive memory for AI agents — FSRS-6 spaced repetition, 29 brain modules, 3D dashboard, single 22MB Rust binary. MCP server for Claude, Cursor, VS Code, Xcode, JetBrains.
Open the project to explore the idea and the code.
Open the project to explore the idea and the code.
The self-improving Agentic QA harness with Memory. Write tests in natural language. Catch regressions before releases ship.
MemoryBear Equip AI with human-like memory capability
The local-first LLM Wiki: open-source knowledge graph builder, RAG knowledge base, and agent memory store. Built on Andrej Karpathy's pattern. An Obsidian alternative for personal knowledge management, AI second brain, and durable Claude Code / Codex / OpenClaw memory.
Using BFS & DFS like techniques over LLMs for iteratively exploring the solution search space at scale.
An AI-assisted coding agent that runs in your terminal and supports many providers and models.
MemSkill: Learning and Evolving Memory Skills for Self-Evolving Agents
Full computer-use for AI agents. Self-learning workflows. Native macOS. No screenshots required.
Remove AI features, telemetry data reporting, sponsored content, product integrations, and other annoyances from web browsers.
Decision audit trail + persistent memory for AI trading agents. Outcome-weighted recall, SHA-256 tamper detection, 17 MCP tools.
Caveman Compression is a semantic compression method for LLM contexts. It removes predictable grammar while preserving the unpredictable, factual content that defines meaning.
Agent skill for elegant and simple html plans, architecture diagrams, or whatever else you can think of.
Open Source Agentic Security Scanner.
Fully recreating Claude Code from scratch, so anyone can learn how it works.
Tutorials, Guides and Agent Skills Directories.