open-source

Open-source runtime action grounding framework for AI agents. It verifies proposed agent actions against live systems of record before execution and can return allow, hold-for-human, or block decisions.

A research project and interactive visualization on how agent-generated code behaves after merge in real-world open source repositories, with a paper, replication package, and data-driven findings.

Self-hosted registry for AI agents and MCP servers that supports capability-based discovery, health checks, and simple JSON/HTTP registration.

A local Anthropic-compatible proxy that lets Claude Code work through ChatGPT, Kimi, Grok, or Cursor subscriptions, with provider routing, diagnostics, and model mapping.

Open-source compiler that turns a SKILL.md into a typed, runnable agent harness. The page explains the architecture, the compile/run flow, and how it aims to make agent behavior more faithful across small, local, or frontier models.

Open-source input scanner for AI agents that strips hidden instructions and scans text, files, images, PDFs, QR codes, audio, and video before they reach an agent.

A macOS terminal app that lets AI agents “live” in a shared environment with the user, with customizable scenes, UI, and behavior packs.

An interactive calculator for AI product unit economics. It models token costs, retries, prompt caching, model routing, infrastructure, support, and target gross margin, and can share scenarios through URL links.

An open-source governance layer and policy-decision daemon for autonomous AI agents, designed to enforce guardrails, budget limits, and runtime policy checks.

Traicer is a developer tool that captures selected AI coding conversations, redacts secrets, encrypts traces, and stores them in seller-controlled S3-compatible storage for later marketplace use.

Open-source social media skills for AI agents that generate platform-native posts, threads, carousels, replies, ad copy, and content calendars in a brand voice.

An open-source CLI for inspecting live LLM inference processes on NVIDIA GPUs. It breaks down VRAM usage by component, shows runtime and model details, and projects memory savings from quantization strategies.

A self-hosted Python runtime for building agentic AI apps with server-driven UI, audit logging, RBAC, sandboxing, and pluggable model orchestration.

A macOS menu bar app that turns AI coding token usage from Claude Code, Codex, and Gemini CLI into a Pokémon companion, while also tracking spend and usage limits from local logs.

Darby is a Hugo documentation theme with built-in in-browser AI Q&A, instant search, dark mode, and polished doc components for technical sites.

A local Rust-plus-Python middleware layer for LLM apps that claims to reduce token usage and add in-process security checks before prompts are sent to a provider.

An open-source tool that watches Playwright failures, inspects the live page, and opens a GitHub PR with a proposed fix.

An open-source multi-model teaching assistant for UIUC ECE 120, with a Gradio UI, retrieval-augmented search, and evaluation scripts. The repo includes setup instructions, a live Hugging Face demo, and details on how it uses separate models for retrieval, generation, moderation, and ranking.

Wisp is an open-source desktop AI overlay that lets you trigger context-aware prompts with a hotkey, capture selection/clipboard/browser/screen context, and stream replies back into your workflow. It includes voice input/output, privacy controls, model-provider flexibility, and MCP support.

Kote is a developer memory layer that captures AI chats, Git activity, and development decisions, then makes them searchable in the IDE, web app, CLI, messaging apps, and MCP-enabled assistants.

Local-first AI trading research assistant for market context, strategy generation, and Backtrader backtests. It supports multiple LLM providers, exposes an OpenAI-compatible API, and keeps runs and data stored locally.

Self-hosted LLM gateway for cost-aware model routing, spend tracking, and policy-based traffic control across multiple providers.

Heckle is a QA copilot for agentic app development. It lets you describe a bug in plain language while you use the app, then packages browser context like DOM, console, network, and path details for a coding agent to fix.

Open Science is an open-source AI workbench for scientists. It combines literature, code, figures, reports, and review into a local-first desktop workflow with model choice, reproducible artifacts, and scientific agent skills.