Developer Tools AI Tools

An Apple Silicon–optimized inference build of Bonsai 1.7B with custom Metal kernels, benchmark results, quick-start instructions, and a bundled OpenAI-compatible server.

Trama is a VS Code tool for agentic coding that puts a map of the app next to the code and makes AI-generated changes go through approved Proposals before anything is built.

A technical Ubicloud blog post about improving tool-call schema accuracy in vLLM for open-weight models, especially Kimi and DeepSeek, using constrained decoding.

Bellman is an MCP-based system for cross-session, cross-provider agent collaboration. The repository describes rooms, role-based join codes, peer confirmation, message passing, room surfaces, and support for MCP plus A2A integration.

An open-source Pac-Man benchmark where AI decision models compete in real time against scripted ghosts, with a public leaderboard, replayable games, and instructions for benchmarking your own endpoint.

A Claude Code plugin that delegates subtasks to Codex, Antigravity, opencode, or a local model, each running in its own tmux session, then returns only the final report.

A React-based package for adding lip-synced Gaussian-splat avatars to AI voice agents, with support for OpenAI Realtime, GPT-Live, Qwen Realtime, ElevenLabs, Vapi, and LiveKit Agents.

MineTrials is a benchmark-style writeup exploring how far AI agents can get in one hour of Minecraft, comparing models and harnesses and highlighting performance, cost, and a best run.

Python CLI wrapper for Jev that analyzes JSON, NDJSON, JSONC, and text logs with typed AI questions and confidence-aware answers.

Release notes page for Claude Code with a long version history and recent updates, including new AGENTS.md support and multiple bug fixes and workflow changes.

A curated GitHub list and benchmark leaderboard for ultra-lightweight, zero-dependency AI engines in C/C++/Rust/Assembly, covering local LLMs, vector search, memory, GEMM, and agent tooling.

A Python library that traces multi-step AI agent workflows and compiles repeated trajectories into deterministic, low-latency code with runtime guards and fallback to the original agent.

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.

A lightweight Markdown variant designed to reduce token usage for LLM consumption using single-character markers and context-aware parsing.

Ramanujan is a terminal-based multi-model agentic workbench for research in computational mathematics. It lets users chat normally, break down problems, survey literature, run parallel subagents, and produce a consolidated report with deterministic checking via SymPy and Z3.

A practical article explaining how to run AI models directly in the browser with Transformers.js, including setup, how it works, browser support, and example use cases like translation and grammar checking.

A governance framework for multi-agent AI systems, with agent lifecycle controls, memory, runtime hooks, and quick-start scaffolding for governed projects.

OpenAI’s model documentation page for the API, with navigation into model selection, pricing, reasoning, text and code, embeddings, moderation, and related developer docs.

Stripe’s llms.txt entry surfaces core docs, API reference, SDKs, webhooks, Checkout, Connect, and a note to always check the latest package version before installing.

A safety-evaluation harness for AI agents operating physical equipment. It validates proposed robot intent against declared hardware and deployment limits, records machine-readable refusal reasons, and includes a sample corpus plus a runnable evaluation script.

A GitHub repository for a custom llama.cpp backend that runs Qwen3-0.6B GGUF models on an Axera AX8850 NPU accelerator, with build/run notes, environment flags, architecture details, and test status.

Conveyor is a software factory for agent-written code. It queues work from requirements, system design docs, and decisions, then lets operators confirm plans while local agents implement and review changes with traceable lineage.

A lightweight SQLite FTS5 memory layer and MCP server for Google Antigravity agents, with CLI tools for storing, searching, and syncing facts.

A DeepSeek Harness plugin that uses an LLM to verify and rank candidate outputs with select, compare, track, and rollout tools.