Open-source components of the Modular Platform: the Mojo language and the MAX framework for AI development and deployment.
Composable skills for coding agents like Claude Code and Codex covering alignment sessions, TDD, debugging, and shared project language.
Native desktop app that configures Logitech mice, keyboards, and webcams locally, as an alternative to Logitech Options+.
A plugin suite that gives AI coding agents a structured development methodology: spec-first design, planning, and TDD via subagents.
Official Cursor plugins for developer tools, frameworks, and SaaS products, each packaged as a standalone directory with a manifest.
An open-source AI job-search pipeline that runs inside AI coding CLIs to evaluate offers, tailor CVs, and track applications.
Self-hosted server that gives AI coding agents shared, persistent memory as git-backed markdown, with cross-agent handoffs and search.
AI tool that turns a topic or keywords into short videos with scripts, voiceover, subtitles, music, and matched or AI-generated footage.
Kubernetes-based agent execution runtime that multiplexes many sandboxed agents onto shared workers with fast suspend/resume.
Desktop app that turns terminal coding agent CLIs into a coordinated multi-agent office run by a supervisor clone of you.
Open-source product analytics platform combining session replay, feature flags, experiments, error tracking, and AI observability.
An Android app, iPhone web app, and Python CLI that turn a Google Maps Timeline export into an animated MP4 travel video on device.
Open-source context database for AI agents that stores knowledge, memories, and skills as a searchable virtual filesystem.
Tooling that makes AI coding agents reply tersely and compresses their input context to cut token usage.
Open-source project management tool for tracking issues, running sprint cycles, and managing product roadmaps across teams.
AI red teaming platform that scans AI infrastructure, MCP servers, agent skills, and models for security vulnerabilities.
Rust vector index with Python bindings using Google's TurboQuant quantizer for memory-efficient local vector search.