Find the open-source solution that fits.

loop-creator helps decide whether recurring work deserves a bounded agent loop, adapts one of 50 patterns, or designs a new loop through a plain-language interview. It can diagnose missing brakes, reward hacking, self-approval, and stale state, with a separate checking instance and explicit stop or human-question conditions.

gwdg-gipplab-skill is a Claude Code plugin and skill guide for the GWDG/NHR-Nord Göttingen clusters Emmy and Grete. It organizes account types, SSH and transfer paths, Slurm jobs and partitions, storage quotas, Lmod/CUDA/Apptainer environments, and Hugging Face caching into focused references. Install it through the plugin marketplace or use the bundled skill when working on those systems.

figure-extractor finds complete figures and images in PDFs, arXiv papers, or HTML, using captions and bounding boxes to avoid cutting off multi-panel content. It crops with PyMuPDF at a chosen DPI and can build a contact sheet for quick visual checking. The project is a shell CLI, not a hosted or in-agent service, so Python, PyMuPDF, and a pip-capable runtime are required.

allincms-content-ops organizes the path from raw sources and a living wiki to search intent, web drafts, media, review, publishing, and indexing. Humans and supported coding agents can share the same file-based workflow, while release gates require traceable sources, complete publishing fields, audits, and an independent adversarial review. Full execution is best suited to environments with browser automation and subagent support.

tower-defense-skill is a native Canvas 2D tower-defense game built around protecting a goddess core. Choose from two maps and three difficulties, build and upgrade six tower types, exploit fire, ice, and lightning matchups, and survive enemy and boss waves. Souls earned during waves unlock hero draws, while deployed heroes can receive a patrol radius; results include progression, achievements, and map-specific rankings.

Litmus closes the gap between changing code and resolving a user’s problem. It drives the running application as the reporting user, follows the exact ticket steps, and inspects the live request and response, frontend state, and database. The tracker is updated only when the defect and its absence can be shown, so a green unit test, backend curl, or database row alone is not treated as proof.

fullstack-deploy takes a natural-language website idea through requirements, Supabase schema and access design, application development, GitHub repository setup, and Vercel deployment. It is aimed at prototypes and independent projects and can work with several Vercel-supported frameworks before handing back a public URL. The workflow still depends on the user’s third-party accounts, permissions, quotas, and regional access.

Run the skill in a website repository to detect the stack, choose separate policies for retrieval and training crawlers, and audit the existing foundations. After approval, it can add or repair robots.txt, llms.txt, sitemap wiring, canonical and meta tags, and Schema.org markup, then verify the deployed files. These signals may improve discovery and attribution, but they do not guarantee AI citations, and robots.txt remains advisory.

geng-skills combines deterministic statistics, duplication checks, and image inspection to flag suspicious data or edited figures in LaTeX or Word manuscripts before submission. The flagged material is then interpreted by an AI workflow, and the project exports a PDF with risk levels, image-forensics notes, an overall conclusion, and a disclaimer. It is a self-check aid, not a substitute for examining the original data and methods.

multi-agent-research routes a question through clarification, methodology and ethics checks, web, document, and academic scouts, source verification, synthesis, gap detection, adversarial review, and report writing. It rates or drops weak sources, preserves contradictions and uncovered areas, and returns a structured report in the language of the question. Seven phases and three quality gates make the workflow explicit, but the quality of the answer still depends on the available sources.

Feed interview-retro a recollection, transcript, or recording link after an interview. It produces either a learning-oriented question breakdown or a personal self-assessment with skill gaps and next actions, choosing the format from context. The default Feishu output needs an authenticated lark-cli setup, while local Markdown works without it.

smooth-scroll-storytelling provides a no-build recipe for quiet editorial pages with elastic scrolling and staged reveals. It combines a proxy-driven GSAP scroll layer, word-level blur reveals, sticky scenes, a drawn SVG guide line, and a scroll-responsive canvas visual, with reduced-motion and no-JavaScript fallbacks. The included skeletons and demo are meant to be reused in native HTML, CSS, and GSAP projects.

Describe the scene you want, and this Claude skill library structures a Higgsfield prompt around model, camera, subject, look, and action. Its sub-skills cover character consistency, model selection, Seedance workflows, Kling motion control, campaign planning, and diagnosis of failed generations. The library writes and refines prompts but does not submit jobs itself; generation still requires a Higgsfield account and the website, CLI, MCP connector, or official execution skills.

Knowledge Panel Playbook guides a person through the web assets used in its own documented panel process: Person JSON-LD, a referenced Wikidata item, a Wikimedia Commons headshot, and consistent bios across platforms. It can generate the schema and supporting files, then print a creation checklist. The repo is available as a Claude skill, pasteable prompt, or templates, but Google controls whether and when a panel appears.

Agentic Learning Studio has Claude write a typed lesson blueprint, then uses a bundled zero-dependency renderer to produce one interactive HTML file. Lessons can include a mental map, sequenced modules, glossary tooltips, examples, and a graded knowledge check, all usable without a server, account, or API key. The same skill can also turn a brief into an installable Agent Skill and requires Node.js 16 or newer.

findable searches current SEO, GEO, and llms.txt guidance, reads the project’s stack, and can apply low-risk changes such as robots.txt, sitemaps, meta tags, JSON-LD, canonical links, and missing alt text. Sensitive edits are recorded in TODO SEO.md with the source URL, exact file, and rationale for you to decide. It works across common static-site and web frameworks, but it relies on current web research.

Point the skill at an application repository to turn its routes, authentication, roles, domain models, and implemented journeys into a role-based flow map. It creates an editable FigJam board when Figma MCP is available, or a self-contained Markdown file with Mermaid otherwise. Planned or uncertain paths are marked separately so the diagram distinguishes the product as built from the product as intended.

b3sty-skill gives Claude Code and Codex a reusable rule set for RedM and FiveM Lua development. It covers resource structure, server authority, networking, NUI, runtime behavior, databases, debugging, and native calls, with separate guidance for each game. The repository also keeps shared and game-specific working memory, making it suited to implementing or reviewing FXServer resources.