Shape agent capabilities into your workflow.

Three.js Awesome Graphics Agent Skills packages reusable implementations for cameras, procedural materials and geometry, water, weather, volumetrics, post-processing, and visual validation. Rather than acting as an API cheat sheet, it gives coding agents concrete patterns and example assets for building sophisticated scenes and games. The pack installs through npm into supported hosts including Codex, Claude Code, and Cursor.

This collection gives coding agents 66 game-development skills plus a router that detects the project engine and the task being requested. Users describe a feature in plain language, and the router loads the relevant engine, discipline, genre, and shipping guidance in sequence. It covers Godot, Unity, Unreal, several web engines, Bevy, pygame, LÖVE, and Roblox, using portable SKILL.md folders written from primary documentation and pinned to stated engine versions.

DSPy Agent Skills helps Claude Code, Codex, and compatible agents build, evaluate, and optimize DSPy programs, from core modules and evaluation to GEPA, RLM, and end-to-end workflows. The pack is validated against the DSPy 3.2.x API and includes runnable examples with offline dry-run modes. Python 3.10 or newer is required, while Deno is needed only for the RLM examples.

Startup Founder Skills packages common company-building tasks into agent workflows for pitch decks, investor research, outreach, product requirements, recruiting, engineering reviews, legal operations, finance, and marketing. A shared startup-context skill supplies the company's stage, product, market, team, metrics, and goals so each task can produce output tailored to the same business.

Create paper figures through two distinct workflows: image generation for architectures, schematics, and graphical abstracts, and local plotting for benchmarks, ablations, heatmaps, or any panel where values must remain exact. The skill includes bilingual figure briefs, scripts, and editable handoff formats. Conceptual image generation requires a supported provider such as OpenAI or Gemini-compatible services.

Give the Claude Code skill a short-film idea and it structures the theme, scene, atmosphere, camera rules, and storyboard into a model-ready prompt. It includes genre templates, failure-to-fix examples, and a browser-based prompt library for Sora, Kling, Veo, Seedance, and similar tools. Some original Mx-Shell materials are educational references and require separate permission for commercial use.

VibeTree gives each AI-assisted task its own Git worktree so several features or fixes can progress in parallel with separate terminal state. Desktop, browser, and mobile access are available, but the main branch is still developing cloud and broader platform capabilities; the README recommends release-v0.1 for a stable desktop-only build.

Spine Animation AI helps Claude turn separated character images and a reference pose into Spine 2D skeletal animation assets. It can position parts with image matching, build skeleton JSON and draw order, generate common motions such as idle or walk, and create interactive HTML5 previews. It can also refine an existing Spine skeleton, but users still need well-prepared artwork and a specific animation goal.

JetBrains/skills collects 129 agent skills from 15 upstream repositories into a flat catalog that is easier to browse and reuse. Each skill preserves its original source link, while included CI can scan changed skills or audit the full repository. The collection is intended for demonstration and education, so critical workflows should still be tested independently.

This collection gives compatible AI agents a staged workflow for film and television development, from screening source novels to analyzing plot structure, evaluating adaptation potential, planning vertical dramas, and drafting scripts. Its 31 skills are separated by creative task, while orchestration skills can split long text, coordinate analysis, and combine results. The package covers short-form vertical series, longer dramas, television, and film, but it is an instruction library for writing and evaluation rather than a video production application.

Exam Cram Coach organizes slides, notes, assignments, and past papers into a chaptered knowledge base and provenance-labeled quiz bank. It adapts tutoring, drills, mistake review, and cheat-sheet preparation to the learner's starting point and available time, while marking material-backed claims, AI supplements, and unsupported questions separately. Basic text and Office ingestion uses standard tooling, but some PDF and visual artifact routes require optional local dependencies.

NSFC Agent Skills offers four independent skills for structuring a proposal, creating research figures and roadmaps, managing literature and citations, and checking current funding policies. Researchers can load only the relevant skill in OpenClaw or another Agent Skills-compatible tool for each stage of the application. The collection is advisory: users must verify all generated material, follow research-integrity requirements, and defer to the official NSFC guidance for the applicable year.

Find skills across project, user, and Claude plugin directories, match them semantically to requests, and inject selected instructions into an OpenCode session. The plugin can restore skill context after compaction and includes helpers for reading files or running skill scripts. OpenCode now has native skill support, so this maintenance-mode plugin is mainly for users who need those extra behaviors.

This repository packages creative production, presentations, video, Google Workspace, advertising analysis, and other workflows as portable SKILL.md modules. Users can install the collection across Claude Code, Copilot, Cursor, Windsurf, and related agents, while some skills still require local tools such as ffmpeg, Node, Python, or credentials for external services.

X2Strategy turns quantitative-finance papers, drafts, reports, or text ideas into structured strategy specifications before generating Backtrader code. It parses multiple formats, extracts indicators, signal logic, execution details and risk controls, validates syntax and framework structure, runs backtests, and diagnoses deviations from reported performance. Ambiguities can be flagged for human review and supplemented with user instructions or verified references. A Python environment and model API key are required, and outputs are not automatically suitable for live trading.

Project Architect is a documentation-first agent skill for converting an idea into an implementation-ready blueprint. It asks structured discovery questions, then produces a specification, technical implementation plan, ordered task list with file targets and acceptance criteria, an optional branding guide, and a single synthesized prompt for a coding agent. Each document is presented for review before the workflow advances. The skill follows the Agent Skills specification and can be installed across Claude Code, Cursor, Codex, Gemini CLI, and many other coding agents.

awp-skill exposes RootNet testnet actions for reading protocol state, managing stake positions and allocations, working with worknets, participating in governance, and watching real-time events. Read-only queries do not require a wallet, while staking and other transactions do. Guided onboarding, transaction links, balance notices, and session recovery support repeated use.

Markdown Exporter works as an Agent Skill, a Dify plugin, or a Python command-line tool. It converts Markdown into formats including DOCX, PPTX, XLSX, PDF, HTML, Jupyter Notebook, CSV, JSON, XML, and language-specific code files, with processing kept local. Custom templates can style Word documents and PowerPoint decks. Slide conversion expects Pandoc-style presentation syntax, while spreadsheet and other table-oriented outputs are driven primarily by Markdown tables.