Agnostos: AI research internship outreach CLI. Finds professors matching your resume, drafts honest cold emails, sends them paced, tracks and drafts replies.
A Wayland desktop drawn as a drafting sheet: a tiling compositor in Rust (Smithay) with no bar. A live title block, lettered margins, windows tiled on the zones and parked in the margins, its own lock and login screens.
The chat decision assistant: before you reply, Jev reads the chat, judges intent and risk, and drafts replies you fill in with one tap. You press send. Android · Windows · macOS · iOS | 聊天辅助决策工具:先看懂对方再回复,发不发由你。
CADCraft: computer-aided design and drafting — an open-source, clean-room AutoCAD-style app in pure Rust
CSS Working Group Editor Drafts
轻量、灵活、易上手的Python剪映草稿生成及导出工具,构建全自动化视频剪辑/混剪流水线。本项目的CapCut版本正于 https://github.com/GuanYixuan/pyCapCut 内开发
Drafts your next release notes as pull requests are merged into master.
Private source preview: native Jianying drafts, isolated editing/export, and standalone Agent Skill.
Eleven free Claude skills that run a LinkedIn account: posts off 21 hook formulas, comments, replies, profile score, weekly plan, and a humanizer that strips the AI fingerprint and scores the draft before it goes out.
开源剪映小助手|剪映API | 扣子插件 | Open-source CapCut automation toolkit to generate & download draft files. | skills
The most advanced ORM of JVM, for both java & kotlin
DraftDepot 稿仓
JSONSchema (draft 2020-12, draft 2019-09, draft-7, draft-6, draft-4) Validation using Go
3,516 curated Figure 1 / teasers from ICLR, ICML, NeurIPS, CVPR, ACL, AAAI (2023-2026), with tier badges and FigureForge retrieval-augmented figure drafting
A fast Java JSON schema validator that supports draft V4, V6, V7, V2019-09 and V2020-12
Play a finished motion-graphics ad in sync with the rough drafts it grew from, and export the breakdown as MP4.
Independent, unofficial CLI to edit CapCut and JianYing (剪映) projects — subtitles, timing, speed, volume, templates, cut long-form to shorts. No API needed, reads draft_content.json and draft_info.json (剪映草稿) directly. Not affiliated with ByteDance.
Open source wireframing tool written in typescript, react and redux.
Turn a page, draft, or SEO export into a sourced fix plan inside ChatGPT.
Free, open-source desktop app for macOS, Windows & Linux to write, ask, check and compare across your own documents. Fact-check a draft or report against PDFs, Word, Excel & PowerPoint, with every answer citing the exact page. Compare hundreds of contracts in one table. Meeting transcription. Works with ChatGPT, Claude or your API key.
A lightweight, flexible, and easy-to-use Python tool for generating CapCut drafts to build automated video editing/remix pipelines! Another similar project: https://github.com/GuanYixuan/pyJianYingDraft
A day 0 tool for getting your app on k8s fast
TLS 1.3 implementation in C (master supports RFC8446 as well as draft-26, -27, -28)
An AI-powered Scientific Writing Workflow: Agent skills for drafting, revising, auditing, and resubmitting Nature-style journal manuscripts.
Claude Code / Codex skills that turn a video into a Chinese narration recap (视频解说): scene detection, ASR, VLM, script, TTS, ffmpeg assembly, optional editable JianYing / CapCut draft export (剪映草稿导出). Local ffmpeg + one MiMo key, no GPU. | 用 Claude Code skills 把视频做成中文解说成片,可选一键导出可编辑剪映草稿。
Instagram marketing skills for Claude Code and Codex: captions, carousels, hooks, hashtag strategy, and a weekly plan. Draft, humanize, and publish via Publora. MIT. Content engineering by Creative Content Crafts.
powerful cad kernel for web
WebTransport implementation based on quic-go (https://datatracker.ietf.org/doc/draft-ietf-webtrans-http3/)
Write research paper and literature review drafts with an open-source Python engine that checks citation DOIs against scholarly databases. Export PDF, Word, or LaTeX.
37 tok/s from Qwen3-Next-80B (48 GB) on a 32 GB laptop, 24 without drafts vs 2.8 for llama.cpp; reading 62 vs 16 tok/s. Faster than llama.cpp on all 13 models compared in Q4_K_M. LLMs bigger than your RAM, pure C, CPU or GPU.
Rewrites AI drafts so they read like a person wrote them — runs locally, keeps the facts
Scientific illustration for Codex and Claude Code: turn research papers into visual drafts using a curated image-and-prompt knowledge base, then reconstruct approved designs as editable PowerPoint or draw.io figures.
Agent skill that mines your repo's history to draft a VISION.md, stress-tests it with hard hypotheticals, and iterates with you on an interactive review board.
Open-source AI-assisted per-application routing for Windows with OpenAI/Ollama drafts and a sing-box TUN data plane.
C++ standards drafts
Deployment kit: Qwen3.8-27B EXL3 3.5bpw target + DFlash2 EXL3 5.0bpw speculative draft — launcher, config, model cards
Local-first official-document, task, drafting, reporting, and controlled-sync system for Web and Electron.
Trace mining for post-training. Filter trajectories, draft verifiers, export SFT and RL data.
[COLMW'26 Spotlight] AdaFlash: Adaptive Speculative Decoding via On-Policy Distilled Diffusion Drafters
Qwen3.8-Flash-Next (NVFP4) on DGX Spark in vLLM: one Spark 43.9 tok/s with our disk-backed n-gram table patch, staged gather and reduced-vocab MTP draft; TP2 SPEED 53.7; TP4 CONTEXT 9M KV pool. Launchers, patch, 40-prompt harness, KV ledger, credits. SGLang TP2 day-0 lane kept.
A self-improving daily content system — pulls from high-quality sources every morning and drafts posts in your own voice. Run by a Claude Code agent.
Open source AI-assisted Instagram publishing for newsrooms - draft, review, and post breaking-news content fast, with a human always in the loop.
A visual workspace for game narratives and NPCs. Build worlds, arrange scenes and characters, edit branching dialogue, review AI drafts, and rehearse stories. Collaborate with your own AI through bilingual Skills.
Draft: A Parametric Tool for Robot Design Exploration
AI Agent prototype for tender and bid document workflows, featuring tender parsing, industry Skills, draft generation, role-based review, and traceable AI adoption.
Mirai S (Qwen3.8-27B-S trellis GGUF) on a 12 GB RTX 4070: 262k q8_0 tiered cache, MTP drafting at every depth, harness-proofing, the layer; receipts.
Grader 的 agent 采用五阶段主循环:perceive→plan→act→observe→respond,由 harness 驱动。模型只决定意图与置信度,路由、工具集、风险等级由代码钉死。act 分只读 ReAct、RAG、人工提案、批量初批。四个写动作不在工具表,只能产提案。批改输出结构化 GradingDraft,逐条绑定 rubric 与段落 hash,保证可重放。
ESPN draft and lineup tools for Codex and other MCP clients, with browser sign-in, approval modes, per-team limits, and server monitoring.
Draft a company ontology from an uploaded table or document, verify it against every row in code, and keep the person's judgement so the next upload of the same file starts from it
Interim AI layer for the Craft apps: connections, keys, providers, client, egui widgets (first draft)
去 AI 味:中文写作的 AI 痕迹清理 skill。合并三家规则(含语料实测),并裁决它们互相矛盾的七处。A Claude skill for stripping AI tells from Chinese drafts.
Native desktop multiplayer tabletop for Magic: The Gathering. Manual tables, draft/sealed/cube, and connected-hub Swiss tournaments.
Multi-agent pipeline that turns a research topic into a draft paper grounded in real literature, with citation hallucination detection and claim-support checking.
Six security outcomes an enterprise must achieve to run AI agents, and the controls that evidence them. A draft for public comment.
Open problems, research agendas, and papers in Math for AI Safety — public from draft stage onward
An agentic playbook for pre-filling the Bangladesh NBR e-Return (etaxnbr.gov.bd) using Opencode (but anything will work) + Playwright MCP: an AI agent walks the return wizard, patches your draft against your own data files, and stops at Save draft — you review, pay, and submit yourself.
An environment where open models work as agents: a first-level harness with routing, skills, memory and sandboxed tools, on an engine that drafts and verifies its own tokens.
Open-source bilingual resume autofill agent for verified profile building, recruitment form completion, attachment preparation, and reviewed drafts.
iOS 键盘:在任何聊天 App 里直接出意图、风险与候选回复——复制消息,键盘上见分晓。Swift / SwiftUI,中英双语。An iOS keyboard that drafts replies inside any chat app.
NaiveRT: a self-contained TP8 inference runtime for the Naive N series on eight Blackwell GPUs with UMMA support. One persistent CUDA graph per rank runs draft, eight-row verify, top-p sampling and cache commit with no host in the loop: 1,700 tokens/s at batch size 1, one-second peaks above 2,000.