Recently Evaluated

A detailed technical write-up about training a 3.8B-parameter language model from scratch, including setup, throughput improvements, ablations, benchmarks, and cost/performance results.

MUON is a local control plane for coordinating coding agents like Claude Code, Codex, Cursor, and OpenCode with shared memory, role-based permissions, and human approval gates.

A web calculator that estimates the electricity use and carbon emissions of AI token usage, with everyday comparisons, example workloads, and a detailed methodology page.

An interactive browser demo that visualizes attention in LLM text generation, letting users hover or tap tokens to inspect which earlier tokens influenced them.

Emailclaw is a local-first AI agent that uses email as the main interface for tasks, with subject-based project isolation and scheduled automation. The repository includes a product overview, core features, how it works, and quick-start install instructions for multiple platforms.

A zero-dependency fetch replacement that lets AI agents automatically handle HTTP 402 payments, with buyer and seller flows, spending caps, and MCP support.

A paper page about FrogNano, a 4B coding agent trained with RL on roughly 1,500 SWE environments using synthetic tasks and no distillation from a larger model.

Open-source security agent framework for cloud, code, and runtime. It ships 45 built-in agents for AWS, GCP, Azure, Kubernetes, GitHub, and GitLab, and lets users create custom agents in a single markdown file.

A detailed write-up of a self-hosted setup for running coding agents on a low-cost Linux VPS, using Tailscale, mosh, tmux, and MCP-based tools to keep sessions alive across laptop and phone.

Verso is an open-source macOS app that orchestrates local AI agents for knowledge work, with app connections, local memory, and support for switching between frontier models.

Open-source text-to-SQL engine tuned to your own database, with configurable pipeline stages, policy checks, and an evaluation harness for measuring performance on real schemas.

A detailed technical article describing PTCRunner’s approach to building a Clojure-like language, runtime, and preludes designed for LLM agents, with emphasis on security, feedback, introspection, and scaling.

Parlel is an open professional network designed for AI agents to read and match against profiles, company pages, and roles on the open web.

Bear. CTXPM is an open-source context package manager for organizing AI project resources like Skills, Rules, Prompts, Memories, and MCP into a maintainable repo workflow.

A capability-based microkernel written in SPARK Ada and formally verified with gnatprove, with a thin C SDK for building services and distribution components.

A Linux desktop app for managing multiple AI helpers with separate roles, memory, provider connections, and isolated computer workspaces.

An open-source tmux plugin that shows Claude Code and Codex activity status, with window indicators plus optional popup and sidebar views.

A security-focused article about a Git configuration attack path affecting multiple AI coding agents, with concrete reproduction details and vendor response notes. It also introduces grith, described as an OS-level security proxy for AI coding agents.

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.

Open-source CLI and deployment engine for shipping open-source LLMs to AWS or GCP from a single command, with scale-to-zero and serverless provisioning inside the user's own cloud account.

Hazzel is a small terminal coding agent for working directly in a codebase. It reads files, edits with diff approval, runs commands with permission, and supports git-aware actions, with multiple model providers and bring-your-own-key setup.

A blog post showing how a Game Boy Advance game was built with GPT-6 Astra and an AI pixel art generator, including prompts, asset generation notes, and a playable demo.

A read-only CLI that inventories AI agents, MCP servers, plugins, and AI extensions on a machine and explains what each one can access.

An open-source AI engineering platform that edits, documents, and verifies KiCad circuit boards from a prompt or brief, with gated stages for spec, architecture, parts, schematic, layout, outputs, firmware, and dev plan.