Tag

Zero-to-CAD 1M

Provides one million executable, human-readable CadQuery construction sequences synthesized by an LLM-in-the-loop—each sample includes renders, STL/STEP exports, precomputed DINOv3 embeddings and a FAISS index. Designed for training and benchmarking text/image→3D and CAD-program generation models (Apache-2.0).

video-use

Automates video editing driven by LLM agents: reads word-level transcripts to propose and execute cuts, remove filler words, auto grade color, burn subtitles, and generate animation overlays. Self-evaluates every cut before showing a preview; aimed at talking-heads, tutorials and interviews.

wigolo

Provides a local-first web-intelligence layer for AI agents: search, fetch, crawl, extract, cache, find-similar and agent-style research without API keys or per-query billing, running as an MCP server, REST daemon, or SDK.

Kimi K2.6

Multimodal agent model for long-horizon coding, image-text understanding, and autonomous task orchestration. Built as a 1T-parameter Mixture-of-Experts with 256K context and native int4 quantization — intended for coding-driven design, persistent background agents, and swarm-style sub-agent workflows.

REA: Reverse Engineer Anything

Connects coding agents to local reverse‑engineering tools so an agent can inspect binaries, apps, and websites and return traced code, evidence, and reconstruction workflows. Supports native (Ghidra/Hopper/IDA), JavaScript/Electron, .NET, APKs, firmware and network captures via an MCP server and CLI.

Archify

Turns plain-English system or process descriptions into polished, themeable architecture, workflow, sequence, data-flow and lifecycle diagrams as a self-contained HTML file, with one-click theme toggle, copy-to-clipboard and export to PNG/JPEG/WebP/SVG (native up-to-4× rasterization).

Qwen3.6-35B-A3B

Open-weight multimodal 35B Qwen3.6 model in Hugging Face Transformers format that supports image/video/text inputs and native long contexts (262,144 tokens). Emphasizes agentic coding and preserved reasoning traces (thinking), uses an MoE-backed architecture and is designed for self-hosting with vLLM/SGLang/KTransformers; requires multi-GPU resources for production.

Browser-based agent workspace and sandboxed app platform that lets organizations run AI agents to perform tasks and build shareable 'Gadgets'. Uses Gatekeepers for fine-grained, zero-trust access control and runs on Cloudflare Workers with multi-LLM support.

SWE-ZERO 12M Trajectories

Provides ~12.29M execution‑free agentic coding trajectories (≈112B tokens) sampled from 122K GitHub PRs to mid‑train code and agent models. Uses bash-only actions (grep, git, sed, etc.) so it scales without Docker; trajectories are unverified and intended for mid-training rather than final SFT.

Cangjie Skill

Turns books, long videos, and podcasts into executable, testable AI agent skills using a structured RIA‑TV++ pipeline. Produces multi-file skill packs (BOOK_OVERVIEW.md, SKILL.md, INDEX.md, DIGEST.md), applies triple verification and pressure tests, and can install skills into Claude Code/Cursor for agent use.

ego-lite

Runs multiple AI agents in parallel inside a single macOS browser, giving each agent an isolated Space that can use your real logins without touching your tabs; controllable via an ego-browser JavaScript skill to perform web automation with fewer tokens and faster task completion.

Open-SWE-Traces

Provides 207k+ LLM-generated agent trajectories of code edits and tool interactions for training and evaluating software-engineering agents. Collected via OpenHands and SWE-agent using Qwen3.5-122B and MiniMax-M2.5, multilingual across nine languages and released under CC BY 4.0.