Start with the job, then find an open-source answer

Research Skills packages Claude Code workflows for medical-imaging literature reviews, PhD proposals, stylized paper slides, and editable academic presentations. Its writing workflows require collecting and verifying sources before drafting, auditing citations, and completing submission checks, while scholar-slides needs an additional Python and Node toolchain.

Equip supported coding agents with guidance for writing Terraform HCL, building modules, developing providers, running tests, and creating Packer machine images for AWS, Azure, or Windows with HCP Packer integration. Skills can be installed individually through the skills CLI or as Claude Code plugins.

sandboxd is a self-hosted engine for building apps with coding agents. A request creates an isolated container, runs an agent against the prompt, and exposes the app at a live preview URL. Teams can use the browser console for creation and file review or integrate the API into their product; idle sandboxes sleep and wake on demand. It primarily targets Linux servers with Docker Compose and Git; macOS via Docker Desktop is best-effort. The project is a 0.x beta using containers rather than VMs, so untrusted multi-tenant deployments need additional hardening.

ARC-AGI-2 contains 1,000 public training tasks and 120 public evaluation tasks built from small colored-integer grids. Solvers infer a transformation from demonstration pairs and must produce every test output with the correct dimensions and cells. The evaluation set is intended for previously unseen testing, so repeatedly tuning against it would compromise the result.

Shuorenhua is a Chinese-first rewriting skill for chats, status updates, READMEs, release notes, forum posts, and longer writing. It identifies the scene and protects facts such as numbers, versions, commands, paths, quotations, names, and responsibility assignments. It then chooses rewrite strength, fixes structural patterns before word substitutions, and performs a fidelity check for facts, terminology, and register. An annotation mode can flag issues without rewriting. It supports several agents and explicitly excludes code, logs, configuration, and command output.

JiwuChat combines direct and group messaging, media and file transfer, multi-device synchronization, WebRTC audio and video calls, and screen sharing in a lightweight cross-platform client. AI assistants from providers such as DeepSeek, Kimi, and Gemini can participate in private or group conversations. A public web experience is available, while users who need their own service can deploy the separate jiwu-chat-core backend with Docker. On macOS, the downloaded package may require manually clearing quarantine because of signing-related warnings.

UPskill is a command-line toolkit for creating standard Agent Skills from task descriptions, examples, existing skills, or recorded agent traces. It can generate tests, compare baseline and skill-assisted performance, refine weak skills from failures, and benchmark multiple models across repeated runs. Evaluations can run locally or through Hugging Face Jobs, making it useful for teams that want measurable evidence before distributing a skill.

Run Library Skills through uvx or npx inside a project and it scans installed dependencies for official agent skills shipped by their library authors. The tool reports status, installs new skills, repairs managed symlinks that point to older package locations, and removes managed links for dependencies that disappeared without touching hand-written skill folders. Because the skills live with the libraries, they update with the installed package version; Windows users can choose copy mode when real symlinks are impractical.

Curses converts speech recognition or keyboard input into captions that can be shown in OBS, VRChat, Twitch chat, or Discord. Streamers can design caption scenes, copy the generated link into an OBS browser source, or let the app configure OBS through the obs-websocket plugin. It also supports switching designs alongside OBS scenes and routing chat messages into captions or text-to-speech. Recognition and voice output depend on the selected browser, cloud, or Windows service, so credentials and availability vary by provider.

Salesforce Skills Library collects executable workflows for building Salesforce applications, including Agentforce agents, Lightning apps, Flow, Apex, SOQL, Lightning Web Components, objects, fields, and permission sets. Agentforce Vibes installs and updates the collection automatically, while compatible tools such as Claude Code, Codex, Cursor, and OpenCode can add it with the skills CLI. Each skill starts with a SKILL.md file and may include scripts, references, or assets. The library evolves quickly, so skills can be renamed, reorganized, or removed between releases.

Rai Pal manages universal mods that work across multiple games built on compatible engines. It scans libraries from providers such as Steam, GOG, and Epic, detects installed games and their engines, then helps install, launch, and update the correct mod version. Its UEVR workflow can locate Unreal Engine titles and start the mod for a selected game. Detection for owned but uninstalled games relies partly on heuristics, so results can be imperfect.

Agent Sessions is a local-first macOS app for browsing and searching histories from Codex, Claude, OpenCode, Cursor Agent, Hermes, OpenClaw, Copilot CLI, and other supported tools. It shows transcripts and images, saves sessions, and can open resume commands in Terminal, iTerm2, or Warp. Session data stays on the Mac with no telemetry or history uploads; macOS 14 or later is required.

SWE-Gym provides 2,400 real software-engineering tasks drawn from 11 Python repositories, together with repository context, executable environments, and test-based verification. Researchers can use it to generate trajectories and train agents or solution verifiers. Reproducing the experiments also involves the published dataset, prebuilt Docker images, and instructions for OpenHands or MoatlessTools.

SWE-smith creates Docker execution environments from GitHub repositories, synthesizes tasks such as file localization and program repair, filters instances that break one or more tests, and generates issue text. The resulting tasks and trajectories can support SWE-agent training or evaluation. It requires Docker and was developed for Ubuntu 22.04; Windows and macOS are not planned targets.

This template combines Tauri 2, the Next.js App Router, Tailwind CSS, and TypeScript with preconfigured Biome, Rust formatting checks, and GitHub Actions. Developers can run the frontend inside a Tauri window and package a statically exported build for release. Because a packaged Tauri app does not include a continuously running Node.js server, the frontend must use static site generation rather than Next.js server-side rendering. The author also notes that Next.js may be more framework than many Tauri apps need.

Token Monitor is a desktop widget for macOS, Windows, and Linux. It reads local session data from tools such as Claude Code, Codex, OpenCode, and Hermes to show token usage, costs, session details, and trends, while separate provider-specific checks report supported account limits. Local mode needs no server; optional self-hosted sync sends aggregate usage between devices without uploading prompts, source code, or raw conversation logs.

Meme Search turns a local meme collection into a searchable web library. You can upload or mount image folders, generate descriptions, edit tags, and retrieve images through semantic or keyword search. Image captioning, embeddings, and search run locally by default, while an OpenAI-compatible vision endpoint can be selected for description generation. The Docker-hosted interface has no built-in authentication, so it should remain on localhost or behind an authenticated proxy or VPN.

This repository is a small collection of reusable skills for AI coding agents, currently centered on web-shader-extractor. The skill guides an agent through identifying WebGL, 2D Canvas, or shader-based effects on a website, separating them from the original framework shell, and porting them into a standalone native JavaScript project. Users can install the entire repository or select only that skill. It is a focused coding workflow rather than a general-purpose web scraper, browser extension, or visual editing application.