Recently Evaluated

A Shopify app that audits how well a store shows up in AI answers and rewrites product catalog content for ChatGPT, Gemini, Perplexity, Claude, Copilot, and Google AI Overviews.

SightDiff is a local visual proof tool for checking what an AI coding agent changed before you commit. It captures baseline screenshots, re-renders your app after changes, and shows a before/after diff sheet with changed and unchanged surfaces flagged.

Browser-local tool that checks AI agent instructions for missing written guardrails against 23 public controls. It supports pasted prompts, local files, and public GitHub files, with CLI and GitHub Action options for recurring checks.

Public benchmark for comparing agent memory systems, with separate textual and coding tracks, submission flow, documentation, and API guidance.

TaskFuel lets AI agents discover and pay for third-party tools from a single prepaid balance, with the provider, price, and result shown in one flow.

An open-source reinforcement learning training loop for LLMs, spanning small CPU-friendly experiments like CartPole to async RLVR runs on GPU clusters. The repo highlights a compact codebase, quickstart commands, task customization, and support for algorithms like REINFORCE, PPO, GRPO, and RLOO.

A small skill that teaches coding agents to explain software with compact visuals like diagrams, trees, diffs, pseudocode, and HTML mockups instead of long prose.

An MCP server for AI agents that stores persistent long-term memory in OKF format and searches it with SQLite FTS5.

Talkform is a voice form and survey tool that turns spoken input into structured entries. The page shows a working flow for recording audio, sending responses, and viewing captured entries in table or card layouts.

A Claude Code status-line tool that estimates per-turn costs, subagent spend, cache rebuilds, and usage limits from local session data.

A technical write-up on using AI to reverse-engineer an AVerMedia Game Capture HD II, recover access to its Linux-based firmware, and build custom control software and an RTSP stream.

A research preprint and dataset on how five frontier LLMs disagree when judging 1,000 real-world fact-check claims, with accompanying corpus, raw results, and code repository.

Federaide is a multimodal multi-agent orchestration harness for Mac, PC, and Android, with peer-to-peer agent routing, terminal-based code execution, voice input/output, Telegram access, scheduled routines, memory, and desktop control.

An out-of-tree vLLM plugin that compresses KV cache with STAR-KV to fit more concurrent requests and longer contexts on the same GPU memory budget.

HyperSAE is a Python package for mechanistic interpretability that trains hyperbolic sparse autoencoders on LLM activations. The page includes installation steps, a quickstart example, benchmark tables, and a short software architecture overview.

Self-hosted voice assistant that can answer aloud, dictate text, run shell commands, control mouse and keyboard, and connect to MCP-enabled services. It can hand off heavier tasks to coding agents and run on local hardware or with hosted providers.

Hallucinote is an AI-assisted music authoring tool for Ableton Live that turns prompts into editable song projects, with round-trip sync, review, and branchable iterations.

CLI and GitHub Action that scans repos, logs, and CI artifacts for leaked LLM reasoning-trace blocks from Anthropic, OpenAI, and Google.

Open-source auditing tool for AI agents that checks whether an agent is doing the job it is supposed to do. The repo shows a guided CLI, plugin-based workflows for Claude Code and Codex, and output in JSON, Markdown, and terminal scorecards.

A minimalist, terminal-native coding agent written in C. It runs as a single native binary, supports local models through llama.cpp and Ollama, and can also connect to OpenAI, Anthropic, OpenRouter, and Codex.

An open-source benchmark for testing whether LLMs can follow long arithmetic derivations without using tools. It compares models on recall-proof inputs, logs digit accuracy, and reports results across multiple endpoints and model families.

Aakit is a developer tool for measuring and tracking assumptions made by coding agents, linking each assumption to supporting code and flagging which ones fail. The repo includes a clear explanation of the idea, prior art, and early experiment results.

Demades lets users turn any topic into a roughly 60-second AI-generated podcast clip with three hosts, without sign-up.

A r/place-style site for coding agents where artworks are generated, previewed locally, and merged into a homepage slot one at a time.