Arnis
Turns geographic map and elevation data into Minecraft Java or Bedrock worlds, with desktop area selection, generation previews, and export.
THE RESOURCE LIBRARY
393 AI tools, models, and projects. Original sources close at hand.
Email updates
Get occasional, source-linked notes on what they do, who they may suit, and what to check before trying them.
Turns geographic map and elevation data into Minecraft Java or Bedrock worlds, with desktop area selection, generation previews, and export.
PrismML’s original Qwen3.6-27B-derived low-bit multimodal family, with 1-bit and ternary weights and local setup instructions under the shared demo’s Bonsai1 guide.
A Cohere Labs 30B-total, 3B-active mixture-of-experts coding model for code generation, agentic software engineering, and terminal tasks, with public BF16 and FP8 Hugging Face weights, Apache 2.0 licensing, and OpenCode or Cohere API try paths listed by official materials.
A Cohere Labs Command A+ model variant with W4A4 quantization, positioned around agentic tool use, multimodal inputs, multilingual work, long context, and Cohere-reported lower hardware requirements.
A newer DeepSeek OCR model release for image/PDF OCR, document-to-Markdown workflows, dynamic resolution, vLLM/Transformers inference, and visual causal flow research.
A DeepSeek model family release positioned around long-context intelligence, reasoning modes, coding benchmarks, and agentic task evaluation.
DeepSeek's 552B-backbone multimodal MoE model with 1M-token context, adjustable reasoning effort, public mixed-precision weights, and reference code for inference, prompt encoding, and evaluation.
A Google DeepMind Gemma 4 model family collection with public checkpoints including Gemma 4 12B, a dense multimodal model Google describes around local agentic workflows, native audio input, and encoder-free vision/audio handling.
A Z.ai flagship text-generation model positioned around 1M-token context, long-horizon coding, agentic engineering, public model weights, API access, and local serving paths.
Z.ai's 753B text-generation model for complex coding and long-horizon agent work, with public weights, an official 1M-token evaluation path, adjustable reasoning effort, and several technical serving options.
A Z.ai 320B-total, 18B-active multimodal mixture-of-experts model for coding, agent workflows, visual input, and long-context work, with a 1M-token context window, public FP8 and BF16 weights, API access, and several technical serving paths.
A 0.9B multimodal OCR model for complex document parsing with several self-hosted or API deployment paths.
A low-bit on-device translation model from AngelSlim, positioned around 33-language offline translation, GGUF access, Android demo use, and 1.25-bit compression.
A Tencent-Hunyuan multilingual translation model family with 1.8B, 7B, and 30B-A3B variants, 33-language support, an AngelSlim 1.25-bit on-device option, IFMTBench, training notes, and several deployment paths.
A Tencent Hy Team 295B-parameter MoE language model for long-context reasoning, coding, tool use, and agent-style workflows, with public Hy3 and Hy3-FP8 weights, Apache-2.0 licensing, and vLLM or SGLang serving notes.
Tencent Hy Team's preview-stage 770B-total, 49B-active Mixture-of-Experts language model for coding, document and analysis work, game development, research, tool use, and long-context tasks, with a 1M-token context window, public BF16 and FP8 weights, and dedicated vLLM or SGLang deployment paths.
A Thinking Machines Lab open-weights multimodal model with text, image, and audio input, adjustable reasoning effort, a 1M-token maximum context window, public BF16 and NVFP4 weights, Tinker fine-tuning access, and deployment recipes for major inference frameworks.
An InternLM model series built to reason across scientific paper pages, molecular and material structures, protein sequences, and physical time-series signals, with current Intern-S2 previews, earlier S1 checkpoints, public weights, and deployment guides.
An NVIDIA Isaac GR00T N1.7 vision-language-action model family for humanoid and generalist robot skills, with a 3B model, post-trained variants, GitHub code, inference and fine-tuning notes, LeRobot-format workflow support, and official robotics developer materials.
A publicly available 8B vision-language interaction model and deployable system that watches a video stream, decides when to speak or stay quiet, and can delegate harder tasks while keeping the stream in view, with time-aligned data, training materials, and vLLM-based setup paths.
Moonshot AI's open-weight native multimodal agentic model with a 2.8T-parameter MoE architecture, 104B activated parameters, a 1M-token context window, public weights, API access, and deployment guidance for long-horizon coding, knowledge work, reasoning, and visual tasks.
Moonshot AI’s multimodal model with public weights, a million-token context, configurable reasoning effort and a documented contract for preserving reasoning and tool-call history.
A multimodal agentic model positioned around long-horizon coding, tool use, autonomous execution, and broader software workflows.
A Moonshot AI coding-focused agentic model built on Kimi-K2.6, with a 256K context length, image and video input notes, preserve-thinking behavior, multi-step tool-call materials, native INT4 quantization, and deployment paths through vLLM, SGLang, KTransformers, and Moonshot API access.
An NVIDIA kinematic motion diffusion model for generating human and humanoid-robot motion from text plus pose, joint, waypoint, and path constraints, with local inference, a timeline demo, motion exports, official docs, and benchmark materials.
A Krea AI open-weight text-to-image model family with an undistilled Raw checkpoint for fine-tuning and LoRA training, an eight-step Turbo checkpoint for inference, official local code, Diffusers, SGLang, ComfyUI, hosted routes, and the Krea 2 Community License.
A community-made distillation LoRA that runs Krea 2 Turbo in four generation steps, with rank-64 weights for Diffusers and ComfyUI and a stock workflow. Its separately hosted demo displayed a runtime error on 6 October 2026.
Compare hosted testing with a 75.2 GB Q4 local route, then inspect Poolside’s published trajectories and documented tool-call limitations before deciding whether this long-context coding model fits.
A ByteDance Research unified multimodal model for image and video understanding, generation, and editing, with model files, demos, inference scripts, Gradio setup, benchmark scripts, and a stated 40GB VRAM inference requirement.
Liquid AI's two-size family of 230M and 350M bidirectional masked-language encoders for fine-tuning classification, token-level tasks, retrieval, reranking, and semantic similarity, with 8,192-token context, downloadable weights, official demos, and reproducible evaluation materials.
Liquid AI's 2.69B-parameter text model for local tool use, multi-step agent workflows, extraction, RAG, and long-context tasks, with a 131,072-token context window plus native, GGUF, ONNX, and MLX formats.
Liquid AI's 230M-parameter LFM2.5 text model for lightweight on-device agentic pipelines, data extraction, and edge inference, with Hugging Face weights, GGUF, ONNX, and MLX variants, 32K context notes, tool-use guidance, and official docs/blog materials.
A compact 350M-parameter model for on-device deployment, with 32K context and export paths for several local runtimes; Liquid AI recommends it for extraction, structured output, and tool use rather than knowledge-heavy or programming tasks.
A Liquid AI hybrid model with 8.3B total parameters, 1.5B active parameters, long-context support, tool-use notes, local formats, and deployment paths for edge and agentic workflows.
Liquid AI's 3.1B vision-language model for on-device image understanding, grounding, object detection, OCR with layout annotations, and tool use, with native, GGUF, ONNX, and MLX model paths plus a browser demo.
An inclusionAI sparse language-model collection led by Ling-3.0-flash, with standard and FP8 checkpoints for coding, tool use, long-context work, and efficiency-focused agent serving.
A finance-enhanced Ling 3.0 Flash model for connected research, source review, calculations, valuation and spreadsheet workflows, with a 256K context window, public BF16 weights, a dedicated benchmark, and hosted access.
An inclusionAI trillion-parameter model for demanding reasoning, coding, tool use, and multi-step agent work, with public weights, publisher-reported evaluations, multi-GPU serving notes, and hosted access paths listed at review time.
An inclusionAI instruct model positioned around faster responses, token efficiency, tool use, multi-step planning, and agent-oriented workloads.
A feed-forward 3D foundation model for streaming scene reconstruction, positioned around geometric consistency, long sequences, and efficient real-time inference.
Robbyant's smaller causal-fast checkpoint for action-conditioned interactive video worlds, with a roughly 6.8 GB DiT package, public inference code and paper, differing official GPU setup notes, and declared CC BY-NC-SA 4.0 terms.
A Meituan LongCat open-weights MoE language model with 1.6T total parameters, about 48B active per token, native 1M-token context, coding and agentic-workflow framing, and base, FP8, and INT8 model pages.
A Lightricks audio-video model family with gated component weights, synchronized sound, text-to-video, image-to-video, multishot generation, local Python and ComfyUI paths, fine-tuning tools, and a separate hosted API.
A series of generative 3D world models from NVIDIA, positioned around explorable scenes, 3D consistency, and world-scale generation workflows.
A JetBrains 12B MoE model family for software-engineering AI workflows, with 2.5B active parameters per token, multiple public checkpoints, 131,072-token context notes, tool-use and agent-workflow framing, deployment examples, and a technical report.
A Xiaomi MiMo model family positioned around multimodal understanding, agentic workflows, long-context use, and Pro variants for harder software and tool-heavy tasks.
An OpenBMB 1B-class language model for local assistants, coding agents, tool-use, reasoning, and long-context workflows, with 131K context, Think / No Think modes, standard LlamaForCausalLM architecture, and deployment paths across Transformers, vLLM, SGLang, Docker, GGUF, Ollama, LM Studio, and MLX.
An OpenBMB 2.52B-parameter language model for local assistants, coding agents, tool use, reasoning, and long-context workflows, with 131K context plus BF16, GGUF, MLX, GPTQ, and DSpark variants.
A MiniMax audio-video generation system with public H3-Base weights for text, image, video, and audio reference workflows, native stereo output, 4-to-15-second clips, local 768p generation, a hybrid 2K path, and a direct hosted API.
A community-maintained MiniMax H3 integration index that maps checkpoints, hardware and VRAM starting points, runtimes, ComfyUI nodes, prompting tools, acceleration routes, and deployment options.
A LightX2V community LoRA project that distills MiniMax H3 into four- and eight-step video-with-audio workflows, with Diffusers and ComfyUI checkpoints, single- and multi-GPU inference code, and a 768p four-step option.
A large MiniMax model focused on agentic work, software engineering, tool use, and complex productivity workflows.
A MiniMax native multimodal model with public ModelScope and Hugging Face pages, 1M context, about 428B total parameters and 23B activated parameters, MiniMax Sparse Attention materials, thinking and non-thinking modes, and deployment paths for Transformers, vLLM, and SGLang.
An OpenMOSS 11B vision-language model family with Realtime, Instruct, and Base checkpoints for continuous video streams, offline multimodal work, and further training, plus public inference and fine-tuning paths.
Meta's downloadable 30B dense model for local agent, coding, tool-use, and text-and-image workflows, with a 131K context window, adjustable reasoning effort, and an official 17 GB GGUF model file for 24 GB-class setups.
An NVIDIA 14B text-generation model from the Nemotron-Labs-Diffusion family, focused on switching between autoregressive, diffusion-style parallel decoding, and self-speculation for project-reported decoding efficiency gains.
Cohere Labs' downloadable 2.4B vision-language model for native-resolution documents, charts, OCR, visual grounding, image questions, and task-specific fine-tuning, with multilingual and multi-image input plus Transformers, NVIDIA AutoModel, Axolotl, and community MLX paths.
A NuMind vision-language model for template-guided JSON extraction and document-to-Markdown conversion, with text and image inputs, optional reasoning, multi-page PDF examples, a hosted demo path, full weights, and GGUF variants.
An NVIDIA 4B content-safety model for classifying prompts, optional images, and model responses against standard or custom policies, with a Hugging Face model card, launch post, released dataset, Transformers, vLLM, SGLang, and NVIDIA NIM paths to inspect.
NVIDIA vision-language base checkpoint for continued training with text, document images, and video frames, with BF16 weights and a shared context ceiling up to one million tokens.
An Ai2 remote-sensing foundation model family for satellite imagery and planetary-scale mapping, with v1.2 Nano, Tiny, Small, and Base models, model weights, training code, and a technical report.
An Ornith AI reasoning-model family for coding and agent workflows, with 9B, 35B-A3B, and 397B checkpoints, quantized serving routes, and a training loop that generates tasks, scaffolds, and solution rollouts.
A specialized text-diffusion model for generating OpenUI interface screens from a component library and plain-language brief, with public weights, OpenUI tooling, vLLM and Transformers paths, function calling, and published benchmark code and raw outputs.
A Qwen language world model for simulating agentic environments, with Apache-2.0 Hugging Face weights, a GitHub repo, AgentWorldBench data, prompts, evaluation/deployment scripts, and domains including MCP, search, terminal, software engineering, Android, web, and OS tasks.
A sparse open-weight multimodal model with 35B total and 3B active parameters, positioned around agentic coding, tool use, long-context work, and software workflows.
Qwen's open-weight 27B dense multimodal model for coding, professional work, research, image and video understanding, and longer agent tasks, with controllable reasoning and a 262K native context window.
Qwen's experimental open-weight preview of the architecture planned for Qwen4, pairing a 125B multimodal main model with 6B active per token, a roughly 360 GB repository, 262K native context, and runtime-specific serving recipes.
A StepFun multimodal MoE model collection with BF16, FP8, NVFP4, and GGUF variants, 256K context notes, tool-use and agent-workflow framing, and deployment paths across vLLM, SGLang, Transformers, and llama.cpp.
A Google Research time-series forecasting foundation model with a public GitHub repo, TimesFM 3.0 and archived checkpoint paths, a PyPI package, BigQuery ML documentation, Apache-2.0 source and weights through 2.5, and the separately named timesfm-non-commercial-license-v1.0 for the 3.0 pretrained weights.
Google DeepMind vision-language encoders with original TIPS and current TIPSv2 materials, focused on patch-text alignment and evaluated spatial awareness.
A Microsoft image-to-3D model with O-Voxel geometry, PBR materials, GLB export, inference examples and training code.
An Arcee AI sparse mixture-of-experts model family presented for multi-turn reasoning and tool use, with long-context handling, quantized variants, and substantial deployment requirements to check in the model card.
A downloadable 3B model built from SmolLM3-3B for formal-logic work, including entailment, semantic parsing, Lean-oriented tasks, procedural reasoning, and rule induction, with merged Transformers weights and five GGUF options for local inference.
A Baidu OCR model and code release for one-shot long-horizon document parsing, with public GitHub, Hugging Face, ModelScope, and arXiv materials, Transformers and SGLang examples, and batch image/PDF inference paths.
A community-made MiniMax H3 derivative that adds a hybrid linear-and-softmax attention branch, eight- and 50-step checkpoints, FP8 single- and multi-GPU inference paths, and the training code behind the conversion.
A promptable 3D detection system for real-world scenes, positioned around text, point, and box prompts for spatial perception workflows.
A small Zyphra mixture-of-experts reasoning model with public weights, 760M active parameters, 8.4B total parameters, deployment notes, and project-reported math and coding evaluations.
Audio8's CPU-oriented ONNX package for compact multilingual text-to-speech and zero-shot voice cloning, with INT8 generation models, local voice registration, and HTTP and OpenAI-compatible speech endpoints.
A compact seven-language speech-recognition model with a Transformers checkpoint, optional hotword boosting, a self-contained CPU ONNX package, and an iPhone demo and Swift SDK for local transcription.
A 1.5B speech model for instruction-guided text-to-speech, content and acoustic editing, paralinguistic changes, speech enhancement, and source separation, with public code, weights, demos, ComfyUI nodes, and fine-tuning materials.
A 3B English and Chinese text-to-speech model for reference-free voice design, reference-guided voice direction, cloning, vocal events, and streaming, with CUDA-focused inference and the provider-declared BreezeBlue Research and Non-Commercial License.
A 2B parameter automatic speech recognition model for audio-in, text-out transcription across 14 languages.
A text-to-speech model with in-script control over pauses, emphasis and mood.
A Python speech library with KittenTTS 2 for reference voice cloning and expressive narration, alongside small legacy ONNX models for CPU voice generation.
A compact 82M-parameter text-to-speech model from hexgrad, with model facts, usage examples, voice materials, samples, a demo Space, and a linked GitHub inference library.
A Liquid AI Hugging Face collection for Japanese-tuned LFMs, grouping LFM2.5-1.2B-JP-202606 and LFM2.5-Audio-1.5B-JP materials for Japanese text, tool use, structured outputs, ASR, TTS, and speech-to-speech workflows.
A robust automatic speech recognition project for messy real-world audio, with code, inference and training paths, Hugging Face weights, a technical report, Voices-in-the-Wild-2M, and project-reported results across difficult acoustic scenarios.
A Xiaomi MiMo speech-recognition model focused on Mandarin, English, Chinese dialects, code-switched speech, noisy audio, songs, and multi-speaker transcription.
An audio-understanding model family for speech, sound, music, captioning, time-aware QA, ASR, and reasoning over real-world audio.
A speech and sound generation model family covering TTS, voice design, spoken dialogue, realtime speech, compact speech generation, and MOSS-SoundEffect-v2.0 text-to-audio materials.
A 0.1B speech generation model covering 20 languages, voice cloning, a repository-recommended ONNX CPU path, browser and CLI use, and released fine-tuning code.
An NVIDIA 600M-parameter multilingual streaming automatic speech recognition model for low-latency voice AI and high-throughput transcription, with NeMo loading and streaming paths, documented language tiers, and performance tables.
An NVIDIA 11B full-duplex conversational speech model with streaming speech understanding and generation, realtime interruption handling, a separate tool-call channel, and official offline and container deployment paths.
A multilingual zero-shot text-to-speech model for more than 600 languages, with voice cloning, text-guided voice design, local inference, training and fine-tuning materials, and public demos.
A real-time full-duplex speech-to-speech conversational model with persona control through role prompts and voice conditioning.
A Japanese-centric text-to-speech system from SB Intuitions, with Japanese and English generation, style transfer, and zero-shot voice generation support.
An on-device multilingual text-to-speech system built around ONNX Runtime, with local inference, 31-language support, expression tags, a local HTTP server with OpenAI-compatible endpoints, and examples across browser, mobile, desktop, and edge runtimes. The maintainers say the repository is archived and development and support have ended.
A speech-language model that aligns speech and text into a single synchronized stream.
A Microsoft Research speech-recognition model compressed for offline CPU inference, with two ready-to-use GGUF components, an official C++ runtime, a CPU demo, and documented x86 AVX2 and Arm NEON results.
A multilingual text-to-speech model with voice design, controllable voice cloning, and streaming support.
A locally runnable music model with 2B and XL 4B variants for full-song generation, reference-guided creation, repainting, accompaniment, stem separation, and lightweight style training.
A framework for generating animatable 3D assets from a single image, with mesh, skeleton, and skinning outputs for downstream animation and simulation workflows.
A 10B research model family for text-to-image generation and instruction-based image editing, with Base, Turbo, Edit, and Edit-Turbo checkpoints, bilingual Chinese-English text rendering, local inference code, and public demos.
A research framework that generates 3D human movement and realistic video together, using paired diffusion branches for human-video generation and training.
The FastVideo realtime video generation and editing platform, with backend and web UI setup, local GPU, B200, Docker, Modal, readiness checks, and mock-backend workflows.
A local image-generation interface built around prompt-focused SDXL workflows, with Windows downloads, Colab access, inpainting, outpainting, image prompts, and presets.
A Meituan LongCat audio-driven avatar video model for single- and multi-person generation, with audio-text-to-video, audio-image-text-to-video, video continuation, model weights, GitHub quickstart, and project-reported evaluation materials.
An NVIDIA Labs infrastructure codebase for long video generation, with LongLive 2.0 NVFP4 and FP8 inference paths, parallel training and inference, multi-shot and image-to-video support, async decoding, LongLive-RAG, model links, docs, and configs.
Mage-Flow documents text-to-image generation and instruction-based image editing in one Microsoft 4B family, with Base, RL-aligned, and four-step Turbo variants. Code is available; confirm checkpoint access before a local run, and check the hosted demo status.
A portrait image-animation framework for live-streaming-style video generation research, with offline and online inference, pretrained weights, a Web UI, and acceleration notes.
Qwen's open-weight image model for text-to-image generation, single- and multi-reference editing, marked-region changes, subject extraction, native transparent RGBA output, and 2K workflows.
An NVIDIA Labs codebase for efficient high-resolution image and video generation, with Sana, Sana-1.5, Sana-Sprint, Sana-Video, Sana-WM world-model work, training and inference pipelines, model zoo links, and ComfyUI and diffusers paths.
An agentic video-generation framework for turning ideas, scripts, or longer narratives into planned video workflows, with a project-based web interface, interactive agent loop and terminal UI, storyboards, shot planning, previews, render checkpoints, and configurable model providers.
A music-to-dance video model and framework that uses a reference image, a full music track, and a dance-style prompt, with separate global keyframe planning and local refinement stages for longer outputs.
An agent memory layer that distills agent-run learnings into Markdown skill files, with cloud and self-host paths, Python and TypeScript SDKs, dashboard/API surfaces, sandbox and disk tools, and cross-framework reuse.
A Python framework for tool-using agents and multi-agent networks. AG2 v1 coordinates agents through a hub and typed channels; older AutoGen-derived classes and patterns remain in the separate AG2 Classic package.
Reusable specialist instruction files, including design roles for image prompts, UI components and visual specifications, with installers and converters for AI coding tools.
A CLI routing layer that helps command-capable agents reach web, social, repository, video, RSS, and search sources through ordered backends. For multi-backend channels, its doctor command reports the active backend. The current main guide describes check-only default setup; the v1.5.0 guide uses --safe for checking without installing core infrastructure.
A draft Agentic Resource Discovery specification for publishing and searching agentic resources across federated registries, with v0.91 ARD entries and ard.json manifests, REST search, trust metadata, a public Apache-2.0 spec repo, and a Hugging Face Discover implementation for Skills, Spaces, and MCP servers.
A self-hosted coding-agent memory server with session capture, keyword recall, optional local or remote embeddings, a viewer, and integrations through hooks, MCP, and REST.
A Python agent framework with ReAct agents, tools, skills, human steering, memory, planning, multi-agent workflows, MCP connections, observability, and deployment paths.
A collection of Python examples and tutorials for building AI systems, including agent building blocks, web context, MCP, model examples, and workflow patterns.
An installable writing skill for drafting headlines, short descriptions, interface copy, LinkedIn posts, and longer marketing pieces, with a second pass for revising prose that sounds formulaic.
A self-built VS Code extension and GitHub Copilot app canvas that reads local AI coding-session logs from several harnesses and turns them into dashboards, practice signals, anti-pattern findings, output views, reusable-skill suggestions, and context-health checks.
An installable skill for AI assistants that searches a regularly updated library of 10,000+ community image prompts, returns a small set of matching examples with sample images, and can adapt a chosen template to a specific brief.
A self-hosted Rust memory service for coding agents, with a Git-backed Markdown wiki, a derived SQLite search index, client-specific capture hooks, MCP access and cross-agent handoffs.
Research-agent skills for evidence review, protocol design, analysis and writing, with scientific figures, schematics and SVG workflows for communicating the work.
A coding-agent skill and drawing kit for interactive technical illustrations, with editable SVG and web output, shared motion and readout models, browser checks and optional beta 3D.
A local-first AI workspace for chatting with documents, using agents, connecting local or cloud model providers, adding MCP tools, building agent flows, and running as desktop, self-hosted, or hosted software.
A scriptable command-line manager for finding, installing, inspecting, auditing, and organizing reusable skills across Codex, Claude Code, Cursor, and other AI coding-agent environments.
A React and TypeScript library for building AI chat interfaces, with message editing, branch controls for alternative replies, and runtime adapters for different backends.
A self-hostable AI bot platform connecting messaging apps to model providers, document retrieval, plugins and tools, with configuration profiles for different conversations.
Meta’s React design system with customizable components, themes and page templates, plus a shared CLI that gives developers and coding agents component references and structured version context.
A local-first operator agent runtime for local or configured cloud models, with browser, file, shell, document, memory, task, skill, MCP, HTTP, and Tauri sidecar paths built around a local control loop and on-disk state.
A continuous local AI-agent loop that uses Claude Code or Codex CLI to coordinate 14 business and engineering roles, carry decisions through one shared consensus file, and expose runs through a local dashboard.
A prompt-as-code library for GPT Image 2 with a categorized case gallery, reusable visual prompt templates, a live browsing site, and an installable agent skill for turning image ideas into structured prompts.
A browser automation framework for AI agents that can navigate websites, click elements, type into pages, use custom tools, and run browser tasks through code or CLI workflows.
An AI-first knowledge base and workspace system with agents, memory, scheduled jobs, and local file-based storage.
An always-on system that records detailed computer activity for vectorless recall, with local storage and a cloud-model privacy tradeoff.
Concise-response skills and a local input-compression proxy for existing coding agents, with recoverable originals, CLI and middleware paths, workload-dependent savings, and documented telemetry controls.
A desktop AI agent client combining model and execution-backend choices, local project and app tools, reusable skills, and mobile task controls.
A game-development template for Claude Code with specialist agent roles across design, programming, technical art and UX, including shaders, visual effects, asset pipelines, menus, workflow commands, and review requirements.
Anthropic's reference blueprint for customer shopping agents and staff merchant agents, with runnable vertical examples, backend contracts, provenance and approval gates, and staged writes that require human approval.
A third-party Claude Code plugin with focused SEO skills, specialist agents, site and page audits, schema and AI-search checks, local reports, and optional integrations for live search, performance, analytics, and backlink data.
An agent video skill with a local captions-or-transcription and timestamped-frame path, plus a Gemini video engine, for supported local coding-agent hosts to ask questions about video URLs or local files.
An economic benchmark and nanobot integration that gives AI agents professional tasks, tracks model and search costs, evaluates submitted artifacts, and shows quality, income, and balance changes in a dashboard.
A local code-intelligence MCP server that turns repository structure into a persistent knowledge graph for symbol search, call tracing, dependency questions, impact analysis, and coding-agent context.
A local codebase knowledge graph for coding agents, with a Rust parsing engine, MCP tools, symbol relationships, call graphs, framework-aware routes, auto-sync, multi-agent setup paths, and project-reported benchmark results.
Turns repositories into persistent code maps that developers and MCP-connected assistants can query to trace calls, imports, dependencies, and other relationships across a codebase.
An agent tool-integration layer with Python and TypeScript SDKs, toolkits, authentication, sessions, triggers, tool search, and workbench features for connecting agents to external services.
A frontend stack for building agent-native applications with chat UI, generative UI, shared state, backend tool rendering, human-in-the-loop workflows, and React or Angular app paths.
A TypeScript SDK for bringing AG-UI-compatible agents into Slack, Microsoft Teams, Discord, Telegram, and WhatsApp, with streaming replies, native interactive components, files, tools, and approval steps inside channel conversations.
A multi-agent deliberation skill for Claude Code, Codex, Gemini CLI, and OpenCode, with independent opening positions, cross-examination, dissent-preserving synthesis, kill criteria, and outcome tracking.
A Python multi-agent automation framework that combines role-based Crews with event-driven Flows for state, routing, persistence, human feedback, tools, memory, and model-provider integrations.
A computer-use stack with Cua Driver for background desktop control, agent-ready sandboxes, benchmarks, SDKs, MCP support, and Cua-S1 Nano and Qwen3.5-4B-based checkpoints for bounded interface decisions.
An Agno public data-agent template for asking questions over company data, with read-only SQL analysis, agent-created views in a separate schema, six context layers, a learning loop, AgentOS, Slack, Docker, Railway, and eval paths.
A tutoring workspace with study materials, Mastery Path, editable learner memory, web and CLI access, and configured image and video generation tools.
A ByteDance long-horizon agent harness for deep research, coding, file work, report generation, skills, sub-agents, memory, sandboxed execution, and message gateways.
A visual platform for building agentic workflows and AI applications with workflow and chatflow builders, model-provider connections, RAG pipelines, tools, APIs, logs, and cloud or self-hosted paths.
A CAMEL-based desktop workspace for single-agent and specialist-workforce tasks across files, browsers and connected services, with visible task activity and separate backend choices.
Design-engineering skills for AI coding agents, with separate paths for building motion, reviewing animations, finding worthwhile opportunities, and naming the effect you want.
A vision-model phone agent using AirPlay mirroring and an ESP32-S3 USB input bridge, with a macOS or Linux host, planner and actor, pointer calibration, and step history. It requires hardware setup and currently supports portrait screens and ASCII typing; model requests include screen images.
A Sentient AGI toolkit for benchmark-driven coding-agent skill discovery and improvement, with support for Claude Code, Codex CLI, OpenCode, OpenHands, Goose, Harbor, local/Docker/remote runs, logs, diffs, and reusable evolved skill folders.
A Microsoft Research 27B multimodal computer-use model for browser tasks that works from screenshots, emits coordinate-grounded actions, and ships with public weights, a Fara harness, and a recommended sandboxed MagenticLite path.
An Apache-2.0 terminal agent, reusable runtime, and evaluation suite with stateful ReAct and coordinated Agent Team modes, task-scoped files, approvals, traces, Docker paths, and model-endpoint choices.
An independent Chrome extension experiment for running an on-device browser agent with Transformers.js, WebGPU, Gemma 4, page RAG, tab tools, and semantic history search.
An early desktop workspace for making browser games with AI, combining game chat, playable previews, build history, assets, configurable model roles, local Blender tools, and static web export.
A Git-native agent framework whose identity, rules, memory, tools, and skills live in version-controlled repository files.
An early Godot editor plugin and companion bridge for AI-assisted scene editing, live node inspection and playtests. The Asset Store marks the current version unstable.
An early-alpha coding-agent skill and local CLI for planning deployment, connecting supported hosting and services on your own accounts, recording checks and tracking unfinished setup.
A Google Colab command-line interface connecting Linux/macOS terminal workflows to remote CPU/GPU/TPU runtimes, with script and notebook execution, file and log retrieval, session control and Drive integration.
A public Agent Skills repository for Google products and technologies, including Google Cloud, with installable skills for Gemini API in Agent Platform, cloud basics, onboarding, authentication, observability, and well-architected guidance.
An open-source local AI agent with desktop, CLI, and API access, repeatable Recipes, delegated subagents, model choice, MCP extensions, usage details, and session export.
A local context graph for coding agents, with deterministic repository maps, symbol and call queries, freshness checks, MCP tools, agent wiring, and optional model-written summaries.
A local-first project knowledge graph for code, documentation, schemas, configuration, and optional media, with extracted-versus-inferred edges, query tools, reusable outputs, and coding-assistant skill paths.
A GreenSock skills repository that helps coding agents write GSAP animation code across core APIs, timelines, ScrollTrigger, plugins, frontend frameworks, and performance-sensitive UI work.
Self-hosted agent execution infrastructure with a shared API for harness selection, task progress, persistent sessions and file outputs, using your configured model providers.
A Google Research framework for recursively refining the prompts, tools, memory and other components around a fixed model, with parallel and sequential experiments, saved harnesses and a credential-free local demo.
A Laude Institute agent microharness whose Bash-based runtime keeps one agent thinking between messages, records its trajectory in inspectable files, and connects the same thought stream to terminal, web, Slack, or Telegram.
A context compression layer with SDK, proxy and coding-agent wrapper paths. It compresses selected request content locally and can retrieve cached originals; provider traffic, plaintext cache storage, operational logs and default content-free beacon reporting have separate documented boundaries.
A Claude Code JSONL trace reader with turn-linked explanations, single-session and multi-file project views, plus documented local setup. Its hosted Hugging Face Space was paused on 4 October 2026; the published code and handling notes remain available.
A Nous Research agent runtime with CLI and TUI use, persistent memory, reusable skills, scheduled automations, messaging gateways, subagents, tool configuration, provider switching, and multiple terminal backends.
An agent memory system designed to help agents learn over time rather than only recall conversation history.
A local-first desktop workspace shared by people and coding agents, with editable file-based memory, apps and model/integration connections that may send content outside the workspace. The project labels its licence modified Apache 2.0; current terms are in the main repository.
An H Company computer-use model family with 0.8B, 4B, 9B and 35B-A3B sizes. The optimized FP8, NVFP4 and Q4 GGUF paths concern 35B-A3B; a separate harness and runtime execute the model's actions.
An unofficial Rust Minecraft reimplementation with JSON game observations, player-action commands and an MCP interface for AI agents and scripts.
Hugging Face describes hf as the command-line entrypoint for working with the Hub, with agent-mode output, predictable command help, skills support, and workflows for Hub tasks such as search, metadata, downloads, uploads, repos, jobs, collections, and inference.
A GitHub-native pull request reviewer with default-branch rules, Action, App and staged web modes. Optional write-capable tasks can open fix PRs after CI failures; that separate flow is off by default and requires explicit opt-in.
A Python diagnostic for running fixed behavioral and governance inspections against language models and AI agents, with provider adapters, repeatable suites, cross-provider judging, manifests, scorecards, and CLI, CI, plugin, or agent-skill paths.
A framework for building and deploying agents across research, documents and creation, including websites and slides with live editing, reusable skills, app integrations, and multi-step work.
A JetBrains Kotlin and Java framework for building AI agents with tools, graph workflows, memory, RAG, MCP, A2A, Agent Client Protocol, tracing, Spring Boot, Ktor, and multiple LLM providers.
A local context layer for data agents that connects SQL warehouse context, semantic definitions, wiki knowledge, CLI search, and MCP tools so agents can reuse reviewable data context.
An agent skill combining recent community discussions, project activity, video transcripts, and web research into source-linked briefs, with source diagnosis and coverage limits. Requires Python 3.12+; available channels depend on setup.
A personal AI harness layer that adds persistent context, memory, skills, structured work, and a local dashboard around a coding assistant, with setup paths for Claude Code, Codex, Cursor, Cline, Gemini CLI, macOS, Linux, and Windows.
A realtime framework for voice, video, and physical AI agents, with Python and Node.js paths, LiveKit room participants, WebRTC clients, telephony support, tools, testing, and deployment options.
A local model-server proxy with on-demand loading, groups or matrix rules for model coexistence, idle unloading and a browser interface.
A cross-platform desktop app for turning documents into an LLM-maintained wiki, with source traceability, graph and hybrid search, local API and MCP access, and a companion agent skill.
A self-hosted model service with compatible APIs, a web interface and selectable backends for language, speech and image generation, with model-specific hardware requirements.
An early open-source local inference server that profiles a computer, recommends model configurations that fit, downloads the selected model, and connects it to existing agent harnesses through local OpenAI- and Anthropic-compatible APIs.
A local model gateway and configuration app for supported coding agents, with provider choices, routing groups, saved profiles, and usage views. It changes agent settings and can discover existing subscription sign-ins. Requests go to the chosen provider, and released builds send a daily user-count event unless disabled.
A TypeScript framework for building AI agents and applications with model routing, workflows, human-in-the-loop steps, memory, tools, MCP servers, evals, and observability.
An open-source knowledge-base agent platform for building RAG-backed Q&A assistants and agent workflows, with document upload and web-source collection, vector/full-text/hybrid retrieval, model-provider connections, MCP/tool use, web embedding, Docker/offline deployment paths, and GPL-3.0 licensing.
A memory layer for AI agents and assistants, with library, self-hosted, platform, SDK, CLI, cookbook, evaluation, and integration paths for persistent context.
A Rust-based memory layer for AI agents that keeps documents, embeddings, search indexes, metadata, and recovery state in one portable .mv2 file, with CLI, Python, Node.js, and framework integration paths.
A curated library of portable agent skills for designers and builders using Codex, Claude, Cursor, Aura Build, Lovable, and similar AI coding agents, with workflows for design prompts, landing pages, screenshots, WebGL, animation, frontend systems, and reusable skill folders.
A Cosmic Stack personal AI agent runtime with permission modes, local structured memory, editable personality files, token budgets, provider choices, skills, scheduled work, background operation, multiple channels, and evidence-gated coding completion.
A proxy-based adaptation framework for personal agents that can turn live conversations into reusable skills, keep cross-session memory, and optionally run cloud-backed LoRA training during inactive windows.
A Microsoft framework for building AI agents and multi-agent workflows in Python and .NET, with a separate public-preview Go SDK plus graph workflows, middleware, context providers, observability, and provider choices.
A Microsoft toolkit for adding policy enforcement, identity, sandboxing, audit records, reliability tooling, and the Agent Control Specification runtime layer around AI agents, with Python, TypeScript, .NET, Rust, and Go surfaces.
A Microsoft Responsible AI evaluation harness for AI agents and LLM apps that turns natural-language requirements into generated test scenarios, runs them against models or traced agents, and writes local artifacts for review.
A beta skills library for coding agents, covering frontend design, shader effects, document layout and generated media alongside mobile and fullstack development.
A document-based multi-agent simulation project that builds personas, runs simulated social activity, and produces reports and agent conversations. Its application requires a model API and Zep Cloud; generated scenarios do not establish real-world outcomes.
A multimodal web agent from Ai2 that can navigate browser tasks from natural-language instructions.
A task collaboration platform for human and AI-agent teams, with shared issues, comments, agent assignments, local coding-tool runtimes, scheduled work, a CLI, desktop clients, and a self-hosted server path.
An agent and product integration platform for auth, OAuth, API proxying, TypeScript functions, MCP tool calling, observability, and external API access.
A lightweight personal AI agent project whose v0.3.5 release adds a native terminal workbench, shared resumable sessions, browser multitasking, clearer automations, and more visible context use.
A lightweight personal agent system that runs agents in containers and connects them to messaging channels, memory, and scheduled jobs.
A Perplexity AI command-line tool for observing supported AI-agent activity on an endpoint, with local rule-based detection, read-only artifact scans, live hooks and plugins, optional blocking on supported pre-action hooks, and portable investigation records.
An NVIDIA AI Blueprint for agentic research workflows, with shallow and deep research modes, citation-backed answers, YAML-configured tools and agents, CLI/web/async job paths, evaluation materials, and deployment assets.
NVIDIA's public catalog of agent skills, framed around NVIDIA-verified skills, skill cards, scanning, signing, product-owned source repositories, and compatibility with the Agent Skills specification.
A self-hosted assistant platform with multiple users and experts, shared knowledge bases, messaging channels, scheduled tasks, workspace memory, and configurable model providers.
A self-hostable AI workspace for connected knowledge, search, agents, and actions, with Lite and Standard deployment modes, multiple model providers, granular sharing, and active v4 admin tooling.
An Alibaba code-review CLI that combines deterministic file selection, bundling, and rule matching with an LLM agent, with local diff and full-file review, resumable sessions, CI examples, coding-agent integrations, and delegation mode.
A computer-use MCP service for AI agents and MCP clients, with macOS, Linux, and Windows paths, Codex, Claude Code, Gemini CLI, opencode setup commands, a Codex skill path, command-call examples, and local validation commands.
A beta personal-agent project with local CLI chat, messaging and app integrations, readable memory, and permission tiers for requests from other people and agents. Its self-hosted messaging tunnel has a reported delivery problem with a fix proposed.
A TypeScript-native framework for coordinating multi-agent work from a goal into a task DAG, with auto task decomposition, parallel execution, plan preview and replay, human approval hooks, MCP tools, provider routing, observability, and sandboxed filesystem defaults.
A self-hosted research workspace that keeps sources and notes together, offers selected-context Chat and retrieval-based Ask, and lets you choose local or cloud AI providers.
A local desktop workbench that connects research projects, AI agents, files, code execution, scientific data connectors, artifacts, previews, and an inspectable activity history.
A shared workspace and cross-platform launcher for collaborating with and managing AI agents, with network development continuing through a separate SDK.
A lightweight framework for multi-agent workflows, with tools, handoffs, guardrails, sessions, tracing, sandbox agents, and realtime voice support.
SDKs for MCP plugin interfaces, including sidebar apps, custom file viewers, searchable mentions, settings, and resource editing in supported hosts.
An early CopilotKit template for a persistent specialist-agent workspace with editable pages, per-Dot computers, and separately configured conversation, voice, and Slack services.
An early CopilotKit work-assistant template for Gmail and calendar briefs, delegated tasks and agent computers, with self-hosted application setup and required CopilotKit Intelligence for threads and memory.
A Hugging Face framework for creating, deploying, and using isolated execution environments for agentic RL training, with Gymnasium-style reset, step, and state APIs, client/server environment access, HTTP and WebSocket protocols, Docker packaging, MCP compatibility, and examples for environment builders.
A self-hosted AI executive assistant with specialist agents for management questions, company-document retrieval, decision memory, scheduled follow-ups, and work-channel integrations.
An OpenHands developer control center for running OpenHands, Claude Code, Codex, Gemini CLI, and custom ACP agents across local, Docker, VM, cloud, and company backends, with conversations, plugins, agent profiles, and scheduled or event-driven automations.
A TinyHumans desktop personal agent with a local Memory Tree and editable Markdown vault, visual workflows, agent orchestration, broad app integrations, optional local models, and managed services for sign-in, model routing, search, and some integration paths.
A multi-agent classroom that can turn a topic or document into slides, quizzes, simulations, and teacher-and-classmate interaction, with hosted and self-hosted paths.
A CopilotKit personal-agent alpha with saved tasks, browser takeover, PDF workflows, and mobile/web interfaces, requiring separate conversation, model, and account-service configuration.
A local dashboard and CLI for running parallel research agents in isolated Git worktrees, tracking experiment branches and evidence, and launching runs across local or remote compute with Claude Code, Codex, or OpenCode.
A search-agent project whose current v2 release provides a 30B checkpoint, evaluation code, and web-search tools, while its retained v1 release includes public training data.
An SEO workspace with keyword, ranking, backlink, and site research tools, an MCP connection for AI agents, reusable research skills, and hosted or self-hosted access.
A skill-management layer for AI agents, with local skill search, quality records, controlled skill evolution, cloud package discovery, and an execution harness.
An OpenCode-powered desktop agent workspace with local skills, connected files and apps, and a separate managed library and MCP Gateway for sharing team capabilities.
An MIT-licensed desktop workspace for running terminal-based coding agents in separate Git worktrees, comparing their changes, reviewing diffs, and monitoring local or remote sessions from desktop and mobile companions.
A Mac-native AI agent harness for Apple Silicon, with local and cloud model options, agents, memory, tools, cryptographic identity, MCP and local API paths, privacy-filter materials, plugin routes, and optional macOS 26 sandbox features.
A JavaScript in-page GUI agent for adding natural-language control to web interfaces, with BYOK model setup, text-based DOM interaction, demo and npm paths, plus optional Chrome extension and MCP server support.
A vectorless, reasoning-based RAG framework for long-document retrieval, tree-structured indexing, traceable document search, and agent context workflows.
A research workflow and coding-agent skill that turns papers and associated code into tested MCP tools and paper-specific agents, with public scientific examples and documented setup requirements.
A self-hosted server and dashboard for coordinating agent teams through companies, goals, roles, issues, heartbeats, budgets, approvals, and persistent activity records.
A coding-agent skill and web player for turning textbook PDFs into narrated interactive lessons and stories into paper-puppet films, with static-book output and optional video or LMS delivery.
Reusable coding-agent instructions for tightening prose, checking claims, reviewing finished work and keeping project instructions maintainable.
A local browser-control server for AI agents, scripts, and automation workflows, with a CLI and HTTP API, persistent Chrome profiles, accessibility-tree snapshots, direct page actions, multi-instance orchestration, and browser audit tools.
A Python framework and ecosystem for real-time voice and multimodal AI agents, with audio/video pipelines, transports, client SDKs, structured flows, and subagent support.
A code-driven pixel-art effects library with an optional Claude Code authoring skill, configurable palettes, sprite-sheet exports and timing data for character attacks.
An MIT-licensed instruction and plugin pack that steers coding agents toward understanding the task, reusing existing code and native features, and writing the smallest solution needed, with rules that tell agents not to simplify away validation, security, data-loss handling, or accessibility.
A realtime voice runtime connecting conversation clients to asynchronous backend-agent work, with independent voice-provider choices, task follow-ups and backend-specific permission controls.
A self-hostable RAG and agent-context platform whose v0.27 line adds Knowledge Compilation, Agentic RAG thinking modes, document and data-source connectors, citations, APIs, and agent workflows.
A CLI and MCP toolkit for AI-assisted app investigation, native decompilation, JavaScript and Electron analysis, and evidence comparison, with separate analysis-tool requirements and explicit unresolved results.
A local-first memory layer for AI agents that stores readable Markdown, uses BM25 and wikilinks by default, and keeps embeddings optional.
A self-hosted code-intelligence layer that combines repository graphs, git history, test impact, code-health findings, architectural decisions, generated guides, MCP tools, and a local dashboard.
A client-neutral cybersecurity workflow pack that routes AI coding agents across reverse engineering, authorized penetration testing, security research, and CTF work, with explicit scope gates, tool discovery, case evidence, and cross-platform routing tests.
An open-source desktop AI coworker that turns email, meetings, notes, and assistant conversations into an editable Markdown knowledge graph, then carries that context into email, browser, coding, app, and background-agent workflows.
A research framework linking world-model predictions, external evaluation, and retained updates, with shared code-repair and GUI workflows, recorded-input evaluation, experiment assets, and run records. Full experiments require task-specific models and compute; reported benchmark gains do not establish general-purpose self-improving AI.
A Rust codebase sensor that maps dependencies, scores five structural signals, enforces architecture rules, and exposes scans and before-and-after session checks to coding agents through MCP.
A Microsoft desktop app that records an on-screen task, turns the captured workflow into an editable intent and ordered steps with GitHub Copilot, and can generate a reusable agent Skill or Automation.
A Microsoft project for improving reusable agent skills without changing target model weights, with a benchmark-driven optimizer and a separate SkillOpt-Sleep path for reviewing sessions, replaying recurring tasks after the live task, and validation-gating proposed changes.
An NVIDIA public scanner for AI agent skills, with CLI scans for Git repositories, URLs, zip files, directories, and single files; static analysis, optional LLM semantic review, OSV dependency lookups, and terminal, JSON, Markdown, or SARIF reports.
A public collection of 17 installable skills for building a creator voice, drafting and formatting social content, planning posts, making visual prompts, and reviewing LinkedIn performance with AI assistants.
Speech To Speech keeps voice detection, transcription, the LLM, and generated speech separate, so builders can change one layer without rebuilding the whole voice pipeline. It supports local speech, local or hosted LLM paths, and an OpenAI Realtime-compatible API.
An OpenAI engineering preview and specification for orchestrating coding agents from project work queues into isolated autonomous implementation runs.
A self-hosted team memory hub that turns conversations, completed work, documents, and code into chat memory, reusable skills, linked wiki pages, and code graphs with ownership and access controls.
A reusable skill for calling TinyFish web search, page extraction, agent tasks, and browser sessions from a coding agent through its CLI.
A coding-agent plugin that learns recurring repository facts, prunes unused tools and redirects apparent failure loops, with host-specific controls and model calls charged through the agent.
A self-hostable agent tracing and regression-testing toolkit that connects failed runs, recorded test cases, replay results and CI checks.
A searchable UI and UX knowledge skill for AI coding assistants, with design-system generation, product-specific style and palette guidance, stack-aware implementation notes, and pre-delivery checks.
A ByteDance GUI-agent desktop app for computer and browser operation, alongside the separate Agent TARS CLI/Web UI stack with MCP tool connections.
A codebase and knowledge-base graph tool for AI coding environments, with plugin paths for Claude Code, Codex, Cursor, Copilot, Gemini CLI, and others, plus search, chat, tours, diff impact views, and an interactive dashboard.
Agent skills and a Python CLI for PC game modding, with engine research, generated assets, in-game checks, and shared field notes. Setup and compatibility depend on the game and workflow; fal generation requires a separate API key.
A Go agent harness with asynchronous tool operations, persisted and forkable sessions, a runner executable, and benchmark tooling for builders designing their own agent runtime.
A self-hosted knowledge platform that combines document ingestion, queries over mounted PostgreSQL data, cited retrieval, an editable ontology, bitemporal fact history, reversible entity review, temporal reasoning, and agent-facing tools.
A coding-agent design kit connecting tokens, component specifications, accessibility guidance, framework adapters, validation gates and critique, with Claude Code, AGENTS.md and MCP integration routes.
A self-hostable AI answering engine for configurable search-style workflows, with local and cloud model providers, SearxNG-backed web search, cited sources, file uploads, and Docker setup paths.
A real-time framework for voice conversations with customizable 3D characters, combining model providers, speech, memory and reaction logic, facial animation, whole-body motion, and browser playback.
A public companion and virtual-character project with browser and desktop paths, real-time chat and voice, Live2D and VRM characters, local or hosted model-provider choices, and experimental modules for memory, vision, tools, and games.
A locally installed interface for character-driven AI chats, with personas, lorebooks, group conversations, prompt controls, extensions, chat import and export, and connections to local or hosted model backends.
An operating system layer for controlling robots and other hardware platforms with natural-language workflows.
A low-cost 3D-printed robotic arm intended for physical AI research and imitation learning.
A motion capture project using ordinary cameras and markerless tracking for movement analysis and animation, with research-grade capture as its stated aim.
Scientific agent skills for analysis, literature research and visualization, including diagram generation with specified labels and relationships through an external image service.
A Hugging Face robotics library for robot-learning workflows, with models, datasets, robot interfaces, training, evaluation, deployment, and v0.6.0 updates across world models, reward models, benchmarks, dataset tooling, and cloud training paths.
A public humanoid robotics workspace around a low-cost 3D-printed robot-learning platform, with hardware files, assembly documentation, runtime tools, simulation/model assets, identification tooling, and training environments.
A Unitree G1 whole-body teleoperation collection with task-specific datasets for household and real-world manipulation scenarios.
An agent skill for making interactive diagrams from a description or codebase, with editable JSON source, structure and layout checks, and portable HTML that others can explore.
A command-line tool that reads local coding-agent usage logs and turns them into unified or source-focused reports for tokens, models, sessions, cache activity, and estimated costs, with JSON output for further analysis.
An HKUDS project for generating command-line harnesses that let AI agents operate software through structured commands, with a seven-phase build method, reusable agent skills, validation and test paths, and a CLI-Hub for installing published harnesses.
An open-source Mac AI workspace, formerly named June, for chat, dictation, meeting notes, image generation, and local agent work with project-described approval gates; the project describes app state as local by default and publishes verification evidence for Clovy API.
An HKUDS coding agent with terminal, desktop and browser clients sharing a local background service, durable sessions and goals, reusable skills, permissions, automations and a Paper2Code research-reproduction path.
A DeepSeek coding-agent workspace with desktop and local Web UI paths, configurable model providers, and replaceable plugins for tools, sessions, permissions, and the agent loop.
A self-hosted sales and support CRM centered on WhatsApp, with AI agents, lead pipelines, human handoff, follow-up rules, and a choice of AI providers. It uses a VPS and Supabase; AI features need a provider key.
An agent skill and local FFmpeg toolset for cutting, captioning, adjusting, checking, and exporting video through structured commands, with a probe-first workflow and optional MCP interface.
A browser-based multi-track video editor with local workspace files, on-device transcription and captions, scene analysis, local voice and music tools, effects, subtitles, and in-browser export through modern Chromium APIs.
A local agentic HTML editor that uses existing coding-agent CLI sessions to turn Markdown, data, and notes into exportable HTML, PNGs, decks, social cards, data reports, and web prototypes with skill templates and sandboxed preview.
A Rust-based terminal tool for running coding agents, with persistent memory, multi-session and swarm workflows, provider switching, local-model paths, reusable skills, MCP support, and cross-platform releases.
A desktop meeting assistant that captures microphone and system audio, transcribes locally with Whisper or Parakeet, and generates summaries through local Ollama or a chosen external provider.
A Xiaomi MiMo terminal-native AI coding assistant built on OpenCode, with public source, docs, install paths, build/plan/compose agents, persistent memory, task tracking, subagents, provider configuration, and release assets.
A local-first desktop design workspace for turning prompts into interactive prototypes, slide decks, marketing pieces, and exportable files through a choice of hosted or local model routes.
A local-first AI design workspace that connects coding-agent CLIs to prototypes, decks, media outputs, design systems, sandboxed previews, and export workflows.
An agent-native slide framework for building React-based decks with coding agents, browser preview, comments, assets, present mode, and HTML/PDF/PPTX export.
An Apple Silicon Mac meeting assistant with local live transcription, local transcript storage, and optional local or cloud-backed retrieval from the user's own notes.
An open-source design editor that opens Figma and Pencil files and gives built-in AI, coding agents, scripts, and custom tools access to the same editable document structure.
An open-source collaborative design platform with an official MCP server that lets AI agents inspect and modify Penpot files, components, tokens, styles, layouts, and assets through hosted or local connection paths.
A Prime Intellect coding and research agent that uses persistent Python, child agents, saved working knowledge, and background sessions to keep larger tasks moving across terminal visits.
A self-hosted AI accounting app for receipts, invoices, PDFs, and transactions, with custom prompts, fields, categories, provider or local LLM options, Docker/PostgreSQL setup, exports, and an early-development warning.
A lightweight AI-native terminal and developer environment built with Tauri, Rust, and React, with a terminal, code editor, file explorer, web preview, AI side panel, BYOK providers, local model support, and approval-style file tools.
A vendor-neutral agent harness with a chat UI, HTTP API, TypeScript SDKs, model and MCP connections, reusable skills, approvals, context management, sandboxed execution, and local or hosted deployment paths.
An agentic development environment born out of the terminal, with built-in coding-agent workflows and support for bringing external CLI agents into developer work.
A Python inference library that streams model weights layer by layer so very large language models can run with far less GPU memory, with PyPI installation, local-model and Hugging Face paths, optional 4-bit or 8-bit compression, CPU support, and an Apple Silicon route.
An Apache-2.0 Python framework that turns existing Transformers or vLLM-served language models into bounded choice, yes-or-no, and score decisions, with levels for zero-label bias correction, calibration, and question-specific decision heads.
An Apple repository for building on-device AI around Core AI, with export recipes for supported open-source models, Python primitives for PyTorch authoring, Swift runtime utilities for macOS and iOS apps, a model catalog, CLI paths, and coding-agent skills.
An AI-assisted enterprise-architecture toolkit for principles, requirements, risk, procurement, design review, diagrams, and traceability, with project scaffolding for Codex and other coding assistants.
A curated collection of DESIGN.md example files inspired by public websites, intended to help AI coding agents understand visual systems, design tokens, layout rules, and UI guardrails.
A Python library for extracting tables from PDFs into CSV or pandas, with controls for table regions and column boundaries and optional machine learning that can use OCR for scans.
An Apache-2.0 contrastive model and local server for typed decisions and candidate ranking, with separate state and action embeddings, reusable candidate caches, a browser playground, public reference heads, and fine-tuning tools.
A preview Cloudflare workspace layer for AI agents, with a Durable Object and SQLite-backed virtual filesystem plus container, Worker shell, and isolated JavaScript execution backends that can operate on the same files.
An incremental data engine for keeping AI-agent and LLM-app context fresh, with Python-native pipelines, delta-only processing, lineage, connectors, and targets for vector, graph, relational, and warehouse stores.
TencentCloud sandbox infrastructure for AI agents, with KVM MicroVM execution, E2B-compatible code workflows, templates, snapshots, an E2B-compatible Volume framework, preview Kubernetes deployment, and preview S3-backed cross-node pause and resume.
A synthetic data framework for related text and image columns from scratch or seed data, with configured model endpoints, dependency-aware generation, validation, and quality scoring.
A format specification and CLI toolkit for describing a design system to coding agents, positioned around persistent visual guidance, linting, and token-level design workflows.
A Python framework for extracting fields, classifying records and summarizing groups across collections of text, with a browser workspace for inspecting sample and intermediate outputs.
A document conversion toolkit with structured text, tables and layout, native chunking for retrieval, Markdown and JSON exports, and local model configuration.
Linux sandboxes for AI-generated code, with Python and JavaScript SDKs, a Code Interpreter, templates and pause/resume workflows through a managed service or public runtime.
An Apache-2.0 streaming inference framework for running sparse mixture-of-experts language models from SSD on Apple Silicon, with two matched 4-bit preview checkpoints, trained routing and LoRA adapters, and an OpenAI-compatible local server.
A local inference system that joins Apple Silicon Macs or Linux machines into a cluster, automatically places and shards models across available hardware, and exposes a dashboard plus several compatible APIs.
A local runtime that shares large mixture-of-experts model inference across GPU memory, system RAM, and CPU compute, with compatible APIs for chat clients and coding agents.
A command-line interface for Workspace services, with repository notes that it is not officially supported by Google and is still changing.
A video rendering framework for HTML-based compositions, positioned around agent-friendly workflows, previewing, and MP4 rendering.
Datalab CPU-based document extraction with PDF text-layer handling, OCR, Markdown and JSON output, layout boxes, and a local viewer for examining recognized text.
An opinionated CLI for very fast on-device transcription with Whisper.
A desktop co-scientist setup built around scientific skills and bring-your-own-key workflows.
A CPU-GPU framework for serving and fine-tuning very large mixture-of-experts language models, with optimized CPU kernels, SGLang integration, LlamaFactory recipes, quantized paths, and model-specific deployment guides.
An early Apache-2.0 model family and Python runtime for bounded choice, score, and yes-or-no decisions, with English, multilingual, and task-specialized checkpoints plus a router that selects between them.
A local vector index for semantic search and personal RAG that reduces stored embeddings through selective recomputation, with Python, CLI, and MCP routes.
A Zig-based headless browser for AI agents and automation, with JavaScript and DOM execution, CDP connections for Puppeteer, Playwright, and Chromedp, local and cloud run paths, a native agent that can save reusable PandaScripts, and MCP support.
A local parser for PDFs, Office files, and images with Markdown, JSON, and text output, complexity detection, several runtimes, bounding boxes, OCR flexibility, and screenshots.
A unified fine-tuning and deployment platform for supported language and vision-language models, with configuration-based CLI workflows, a WebUI, and supervised and preference-training approaches.
An open-source model gateway for AI agents and apps, positioned around cost-aware routing, fallbacks, provider control, and self-hosted workflows.
A document converter with OCR, Markdown and structured block outputs, page coordinates for checking source layout, and chunks output with full HTML for each top-level block.
A Python tool for converting Office files, PDFs and other inputs to Markdown, with format-specific dependencies, optional vision OCR and a separate MCP server package.
Connects AI assistants to the Unity Editor for scene objects, materials, assets and C# scripts, with optional tool groups for animation, testing, visual effects and other Editor work.
Distributed local and self-hosted inference infrastructure that pools GPUs and memory across machines behind one OpenAI-compatible endpoint, with peer routing, package-backed model splits, and private or public mesh paths.
A document parsing toolkit with PDF and Office conversion, a local library for search and page or block reading, and structured outputs for AI applications.
The official MiniMax CLI for terminal and agent workflows, with documented commands for text, image, video, speech, vision, and search.
A native-PyTorch learning project for training tiny language models from scratch, with public data and checkpoints, tokenizer, pretraining and alignment stages, LoRA, tool-use and agentic-RL examples, plus local inference paths.
A developer platform that combines MAX for serving and building AI models with Mojo for writing compiled, hardware-aware code and GPU kernels.
A Rust interpreter for AI-written Python, with explicit host-function access, resumable execution state, and Python, JavaScript/TypeScript and Rust integration paths.
A local AI server stack that installs and connects model inference, chat, agents, workflows, RAG, voice, image generation, service management, and privacy tooling across Linux, Windows, and Apple Silicon.
A focused Python and Streamlit workflow for using Ollama vision models to extract text and structured output from images or PDFs, with preprocessing, batch runs, custom prompts, and multiple output formats.
An Ai2 toolkit for converting PDFs, PNGs, and JPEG document images into clean Markdown or text for downstream AI workflows, with local GPU, remote vLLM/OpenAI-compatible server, Docker, benchmark, model-card, and demo paths.
Codex plugin examples for reusable workflows, including Figma screens and design files, plus videos with audio and captions.
A privacy-filtering model and local toolkit for detecting and masking personally identifiable information in text, positioned around high-throughput sanitization workflows.
A PDF toolkit that combines deterministic local extraction to Markdown, JSON, HTML, text, or annotated PDF with optional AI handling for complex pages, Tagged PDF structure reading, and automated Tagged PDF output.
An early open-source server for typed, probability-based yes-or-no, choice, and score decisions, using DiffusionGemma behind a Jev-compatible API for routing, filtering, and scoring steps.
General-purpose sandbox runtime infrastructure for AI applications, with multi-language SDKs, unified sandbox APIs, Docker and Kubernetes runtimes, CLI and MCP paths, code-interpreter support, browser and desktop examples, and network-control features for agent workloads.
A Hugging Face toolkit for exporting, quantizing, compressing, and running Hub models through OpenVINO on Intel CPUs, Arc GPUs, and Core Ultra NPUs, with v2.0 making OpenVINO and NNCF the default path and release notes covering newer text, vision-language, speech, video, and diffusion model support.
A document AI toolkit for OCR, document parsing, structured Markdown and JSON outputs, PaddleOCR-VL document parsing, PP-StructureV3 conversion, PP-OCRv6 scene OCR, and workflows that feed RAG or agent systems.
A local Rust library for classifying PDFs, extracting native text into Markdown, and identifying individual pages that may need OCR, with Rust, Node.js, Python, CLI, and browser WebAssembly paths.
A local GLM-5.3-Flash engine with GPU, RAM, and SSD expert tiers, browser chat, a hardware monitor, and OpenAI- and Anthropic-compatible APIs. Windows and AMD support are experimental.
A local animation workspace that analyzes a reference video, presents a storyboard for approval, and uses Claude Code or Codex agents to build and review 2D shots before MP4 assembly.
A self-hosted inference server and production cluster for serving retrieval, document conversion, structured output, content-safety, and generation models behind agents through one API, with on-demand loading, SDKs, integrations, and Kubernetes operations tooling.
A CLI and MCP toolkit that ingests documentation and other sources, structures them, and packages outputs for AI skills, RAG systems, vector stores, and coding assistants.
A Python CLI for fine-tuning and post-training language models from one YAML configuration, with data tools, LoRA and preference-training paths, low-VRAM layer streaming, evaluation gates, export, and serving commands.
Code-execution infrastructure with Firecracker workspaces, HTTP command sessions and separate workspace lifecycles for keeping, pausing, suspending or deleting environments.
A toolkit for spec-driven development, with structured implementation workflows and AI coding agent integrations.
A Datalab document OCR and analysis toolkit for OCR, layout analysis, reading order, table recognition, math-aware output, vLLM or llama.cpp backends, and structured document results.
A browser-based distributed inference project that divides model layers across devices, runs them with WebGPU, and passes intermediate activations over direct WebRTC connections, with hosted and local static-server paths.
An installed Rust node for local and peer-to-peer LLM inference, with model sharding, public or private discovery paths, a web dashboard, OpenAI- and Anthropic-compatible APIs, MCP tools, and optional provider routing.
A native .NET inference engine for GGUF models, with console and browser interfaces, OpenAI- and Ollama-compatible APIs, CPU and GPU backends, multimodal model support, and optional distributed or agent-tool paths.
An inference and serving framework for text, image, audio, video, diffusion, and action models, with heterogeneous pipelines, streaming, and distributed execution.
A local-first voice synthesis studio for voice cloning, speech generation, effects, and voice-powered app workflows.
A local-first desktop studio for voice cloning and design, dubbing, dictation, transcription, long-form audio, batch generation, and app integrations.
A Rust-based engine for extracting text, tables and metadata from mixed files, with language bindings, CLI and server interfaces, configurable OCR and retrieval preparation.
An experimental graph-first programming language for agents, where reviewable .0 source stays the durable artifact while compiler-derived ProgramGraph facts, graph hashes, node IDs, diagnostics, checked edits, and version-matched skills give coding agents a more structured interface.
A living-world benchmark for multi-day, multimodal coworker agents, spanning 100 tasks across professional domains and real tool environments.
A manually curated benchmark for general reasoning in LLMs, designed around high difficulty, broad task diversity, K-12-scope knowledge, and hybrid scoring.
A benchmark for evaluating latent action representations, with pipelines for action semantics, robotic control regression, and broader vision-to-action alignment.
An OpenAI evaluation-data release for studying monitorability, with public eval splits, prompt templates, dataset mappings, and metric code from the Monitoring Monitorability paper.
A benchmark for evaluating how well OCR systems convert PDFs into useful markdown while preserving structure.
A public results dataset for the Open Agent Leaderboard, with agent, model, benchmark, score, completion, error, action-count, and cost fields linked to the Exgentic evaluation framework and paper.
A document parsing benchmark for AI-agent workflows, focused on whether parsed PDFs preserve structure and meaning for downstream evaluation.
A terminal-agent benchmark for evaluating AI agents on hard containerized command-line tasks, with Harbor run commands, task-level registry pages, GitHub and Hugging Face materials, docs, and paper links.
An OpenBMB supervised fine-tuning dataset with 15,036,178 thinking and non-thinking samples across math, code, knowledge, Chinese, instruction-following, and multilingual configurations, used in MiniCPM5-1B-SFT post-training.
A multilingual African-language speech collection with separate ASR and TTS datasets, language-specific subsets, and provider-dependent terms.
Try fewer words or explore all categories.
A starting point for your own inspection. Inclusion is not an endorsement. Check current setup, terms, and privacy at the original source.