Open Science Desktop β local-first, model-agnostic AI research workbench for macOS, Windows & Linux. Open-source Claude Science desktop alternative built on Tauri + MCP + agent skills.
The open-source AI research workbench for scientific research and agent workflows. Local-first, model-agnostic desktop app with extensible skills, MCP tools and connectors, Python/R execution and traceable artifacts for reproducible research on macOS, Windows and Linux.
Autonomous five-seat AI science rig: reproducible exoplanet transit detection on NASA Exoplanet Archive data, with Solana mainnet provenance anchoring.
Classic grid bot: Extended / RISEx / Decibel / N1 / Phoenix
ReviewBench is an open, reproducible benchmark for evaluating AI code review systems on real-world pull requests.
Persistent AI companions on Box, powered by Pi. Simple web, durable execution, reproducible local development.
Local-first visual generation runtime and studio for people and coding agents, with reproducible image and video workflows across multiple providers.
A reproducible local LLM benchmarking platform with versioned datasets, deterministic scoring, and auditable reports.
Local-first MCP App for scientific figures. Import, review, and publish a global library on disk; reuse exact templates in Pi, DeepSeek Harness (dsh), Claude, Codex, Cursor, and Wisp.
Rebuild the JavaScript GitHub Actions your workflows pin and check that their committed dist matches the source.
The AI analyst that shows its work. Runs Python and R on your own machine, and traces every chart back to the code behind it.
Script terminal sessions in TypeScript, render reproducible MP4/GIF/SVG/HTML with Bun
Scibly is an open-source, AI-native learning platform. Turn your existing knowledge into interactive learning experiences.
Explore data center power continuity, energy ledgers & infrastructure assumptions through reproducible synthetic models.
A platform for structuring AI research and studying physics of AI
TypeScript execution journal for retryable tools: SQLite durability, lease fencing, explicit uncertainty, and reproducible crash tests.