Top Picks

BoundaryBench is an open-source benchmark for coding agents running under hardened sandbox policies. It compares harnesses like Claude Code, Codex, Terminus 2, and Grok on Terminal-Bench tasks and includes quickstart, policy controls, and result export tools.

MCP server that lets AI agents control labgrid-managed embedded hardware labs through 47 tools, including power, console, flashing, muxing, SSH, and place management.

Open-source, local-first audit trail and detection layer for AI coding agents. It records tool calls from Cursor, Claude Code, Codex, and MCP servers, adds MITRE ATT&CK tags, correlates risky sequences, and can forward events to SIEM tools like Loki, Elastic, and Splunk.

A local-first Node.js utility that redacts emails, phone numbers, paths, secrets, URLs, and other identifiers from LLM prompts, then rehydrates responses locally using session-based mappings.

An AI incident copilot that watches your screen during an incident, extracts what happened, and answers follow-up questions using the Grafana LGTM stack.

A privacy-first personal finance app for Canada and the USA with budgeting, net worth tracking, goals, bill tracking, investment monitoring, and an official MCP integration.

An interactive browser city that visualizes how an LLM works, with live transformer mechanics such as tokenization, attention, KV cache, residual stream, and sampling.

Free hands-on labs for prompt injection and LLM security training, with real models, a context trace view of assembled prompts, and a multi-level Gauntlet challenge.

A curated-taste API for architecture, art, design, and photography, with both REST endpoints and a native MCP server for AI agents. It offers search, similarity, provenance, and taste metadata across 2,500+ hand-selected projects, plus free keys and documented pricing.

DeepSQL is a self-hostable AI DBA agent for Postgres and MySQL that monitors workloads, rewrites slow queries, recommends indexes and materialized views, and connects through CLI, MCP, and Slack.

A fast, private, self-hostable AI chat app for OpenRouter. It supports bring-your-own-key usage, model selection across many providers, attachments, web search, portable history, and both hosted and self-hosted deployment.

LLMIntel is a runtime telemetry and cost dashboard for LLM apps. The live demo shows spend by model, API key, environment, and tag, plus at-risk spend for retiring models and optimization suggestions.

Shellular is a mobile remote-development app for running AI coding agents like Claude Code, Codex, Cursor, and OpenCode from a phone. It provides access to a real terminal, files, Git, ports, localhost apps, and browser DevTools on the host machine, with end-to-end encrypted relay access and app downloads for iOS and A

Self-hosted AI security testing platform for authorized pentesting and bug bounty workflows, with a local web UI, live agent telemetry, verified findings, and branded PDF reports.

WakaWiki is a Rust CLI that generates and keeps agent-facing documentation up to date for a codebase. It supports multiple model providers, a zero-LLM scan mode, incremental updates, and AGENTS.md integration, with CI-friendly GitHub Actions usage documented in the README.

A native C++/SDL3 desktop IDE built to be driven by LLM agents, with editor, git, merge, terminal, search, and full DAP debugging workflows.

A free public API for LLM pricing, specs, benchmarks, and cost calculations across 1,000+ models. It offers no-auth endpoints, CORS support, an OpenAPI spec, and a LiteLLM-compatible cost map that refreshes daily.

A self-hosted email client on Cloudflare Workers with an AI agent for reading inboxes, searching conversations, and drafting replies. The project includes deployment and setup instructions, feature details, and a clear Cloudflare-based architecture.

OpenScience is an open-source AI workbench for scientific research. It can read papers, plan hypotheses, write and run code, manage experiments, query scientific databases, and present results in a browser-based workspace.

Open-source Chromium fork that aims to keep browser automation and scrapers from being blocked by patching fingerprint surfaces inside the engine. It connects over CDP and is positioned as a drop-in replacement for Playwright, Puppeteer, and other CDP clients.

A transparent MCP proxy and terminal UI for watching real client-to-server tool calls, inspecting frames, replaying captured calls, and debugging live JSON-RPC traffic.

Self-hosted LLM proxy that redacts personal data before requests reach model providers, including inside tool-call arguments and multimodal content. It supports OpenAI-compatible clients, runs locally with zero telemetry, and uses Presidio plus an optional privacy filter for broader PII and secret detection.

OpenWiki is a CLI for generating and keeping agent-oriented documentation up to date for a codebase. The repo shows installation, init and update commands, interactive and one-shot usage, and support for multiple model providers.

An MCP server that gives AI agents access to 8,500+ curated awesome lists and 1.3M+ resources, with hosted and local setup options plus tool docs and example queries.