open-source

Research code for improving agent harnesses through regularized recursive self-improvement, with evaluation, selection, pruning, and domain-specific adapters.

ToolHub is a self-hosted open-source platform that turns scripts into AI tools and organizes them in a hierarchical tree for agents to browse and execute. The page shows architecture notes, execution flow, screenshots, and a live demo/web client reference.

Open-source ontology layer for AI agents and organizations. Anchor builds a local-first semantic model above existing data sources, with a CLI and MCP interface for querying confirmed organizational knowledge.

A local-first desktop AI agent with persistent memory, a browser-capable control UI, onboarding flow, and P2P skill sharing.

ContextVeil is an open-source tool that redacts enrolled secret values locally before coding agents send text to an LLM. It keeps files and commands usable while replacing exact secret matches with placeholders.

Swobu is an open-source LLM switchboard that routes agent requests across providers, regions, cloud accounts, and local models behind a stable model name. It supports fallback routing, protocol translation, and agent setup via CLI or Cockpit.

AgentsDock is an IDE for agentic AI research that lets users connect Claude Code, Codex, and Cursor to their own machine and access them from desktop or mobile. The page shows setup steps, supported tools, and a clear workflow for remote agent research and model training.

Open-source LLM uncertainty and hallucination detection tool that estimates epistemic risk in microseconds using exact-match normalized entropy, with Rust gateway and Python package support.

BiNeuron is an open-source local AI assistant for code generation, file analysis, and document/OCR handling. It detects programming languages, selects models based on hardware, and can run offline with proxy support when needed.

Bestie is an open-source native terminal coding agent harness focused on visibility and control, with sandboxing, provider support, session save/load, and cross-platform installs.

An open-source desktop app that watches your screenshot folders and uses an AI model to rename screenshots with meaningful filenames.

Loopback MITM proxy for Claude Code on Amazon Bedrock that can sniff, hold, edit, forward, or drop requests. The repo includes prerequisites, a quick start, flags, cost notes, and behavior limits.

An open-source Redash fork that adds AI-assisted analytics features such as text-to-query, auto-generated visualizations, alert suggestions, and dashboard creation across several AI providers.

Codexify is a Rust-based local MCP bridge for ChatGPT Web Pro that exposes tools like file access, shell commands, git operations, and project management on your machine. It also supports OpenAI’s native Secure MCP Tunnel and can aggregate other MCP servers.

A terminal-based agentic coding harness for engineers, optimized for z.ai GLM models and built to hot-swap across multiple LLM providers mid-session.

Open Executive is an open-source AI system that acts as a virtual executive team, routing requests through a single executive persona backed by eight specialist Claude agents. The repo includes architecture notes, tech stack details, repo layout, and quick-start steps for running the FastAPI backend and Next.js UI.

Open-source agent harness that starts from a tiny loop, gives the model one exec tool, and stores the agent’s state and sessions inside its own directory. It includes setup commands, model configuration notes, and design rationale.

Personal site for an engineer writing about AI and dev tooling, with visible open-source projects for agent orchestration and data pipelines. The page highlights LLAssembly, described as code-driven orchestration for sub-agents and tool-call workflows, plus other related projects and writing.

Open-source local-first contract layer and runtime for canonical data, retrieval, durable execution, and agent-facing AI. The repo includes a quick start, capability boundaries, client/runtime notes, and examples for running locally and inspecting providers.

Open-source BYOK telemetry for tracking how a brand appears in AI assistant answers. It generates topic-based prompts, runs them across multiple models with your own keys, detects mentions, and charts visibility, sentiment, and share of voice over time.

Pacific Slate is a self-hosted, model-agnostic multi-agent AI assistant with background work, long-term memory, source tracing, and an interface built around cards and a demo monitor. The page also explains the architecture, agent roles, storage boundaries, and how it connects through MCP and an OpenAI-compatible endpo

Experimental page-aware Metal runtime and reproducibility harness for running oversized sparse mixture-of-experts models on Apple Silicon. The repo documents GPT-OSS 120B support, performance results on 64 GiB and 16 GiB machines, and the runtime/harness structure.

Open-source, self-hostable platform for shipping Claude Code-style AI agents to Kubernetes or local Docker Compose, with Slack as the first supported channel. The README shows a command-line workflow, deployment tiers, and support for traces, evals, budgets, and model choices.

A Windows tray widget that shows Claude Code and Codex rate-limit usage, reset times, and burn pace. It runs locally, uses the user’s own Claude OAuth token and Codex CLI data, and includes install and build instructions.