Instruccions per a IA
Disseny d’instruccions i plantilles per a models d’IA
ralph-loop
belumume
Activate autonomous Ralph Wiggum loop mode for iterative task completion. Use when you have a well-defined task with clear completion criteria that benefits from persistent, autonomous execution.
iot-homeassistant-best-practices
jasonraimondi
Home Assistant automation, helper, script, and device control best practices. Enforces native HA constructs over Jinja2 templates and safe refactoring workflows. TRIGGER WHEN: - Creating or editing HA automations, scripts, or scenes - Choosing between template sensors and built-in helpers - Writing triggers, conditions, waits, or selecting automation modes - Setting up Zigbee button/remote automations (ZHA or Zigbee2MQTT) - Renaming entities or migrating device_id to entity_id - Agent uses Jinja2 templates where native constructs exist - Agent uses device_id instead of entity_id - Agent picks wrong automation mode (e.g., single for motion lights)
continuous-learning
mhylle
Pattern extraction framework for learning from sessions. This skill captures valuable patterns discovered during work sessions including error resolutions, user corrections, workarounds, debugging techniques, and project-specific patterns. Triggers on Stop hook, end of session, "save learnings", "extract patterns", or when explicitly asked to capture knowledge from the current session. Outputs learned skills to ~/.claude/skills/learned/.
media-generation
vargHQ
Generate AI videos using varg SDK. Use for videos, animations, talking characters, slideshows, social content. JSX-based - describe scenes, render videos.
team-brainstorm
mhylle
Deep brainstorming using agent teams for adversarial multi-perspective analysis. Use when users want thorough idea exploration with real debate between independent perspectives. Triggers on "team brainstorm", "deep brainstorm", "brainstorm with team", "debate this idea", or when the user explicitly requests team-based analysis. Higher token cost but significantly deeper analysis than single-agent brainstorm.
docx-template-filling
belumume
Fill DOCX template forms preserving 100% original structure - logos, footers, styles, metadata. Zero-artifact insertion. Output indistinguishable from manual entry.
forge
jwiegley
Multi-phase, multi-model deep analysis workflow for complex problems. This skill should be used when the user wants rigorous, multi-model collaborative analysis: deep research with Opus and PAL MCP consensus (GPT-5.2-Pro + Gemini 3 Pro), strategic planning, Sonnet execution with tests, comprehensive review, and adversarial devil's advocate critique. Invoke explicitly with /forge.
rlm-orchestrator
belumume
Implement RLM-style (Recursive Language Model) orchestration for complex tasks. This skill should be used when facing large context requirements, multi-part tasks that would benefit from parallel execution, or when context rot is a concern. Automatically decomposes tasks, spawns parallel subagents, aggregates results, and iterates until completion. Inspired by the RLM research paper (arXiv:2512.24601).
prompt-generator
mhylle
Generate implementation prompts for phase-based project execution using an orchestration pattern. Use when users request prompts for implementing project phases, need orchestration prompts for multi-step implementations, ask for "/prompt", "generate prompt", "create implementation prompt", or want to create structured implementation instructions for subagent delegation. Outputs the prompt to chat and saves to docs/prompts folder.
clawra-selfie
SumeLabs
Edit Clawra's reference image with Grok Imagine (xAI Aurora) and send selfies to messaging channels via OpenClaw
chuinb
yizhiyanhua-ai
This skill transforms users into confident industry insiders through immersive, research-driven learning experiences. It should be used when users want to quickly master an unfamiliar industry, field, skill, or domain — whether for professional networking, investment decisions, career transitions, or pure curiosity. The skill combines Feynman Technique, First Principles thinking, and 80/20 analysis with deep web research to generate beautifully formatted, media-rich learning notes with interactive flashcards and quizzes. Triggers include: "help me understand [industry]", "I need to learn about [field] quickly", "make me an insider in [domain]", "行业速成", "快速掌握", "深度学习笔记", "/chuinb", "/master", or any request to rapidly acquire domain expertise.
gitnexus-cli
docxology
"Use when the user needs to run GitNexus CLI commands like analyze/index a repo, check status, clean the index, generate a wiki, or list indexed repos. Examples: \"Index this repo\", \"Reanalyze the codebase\", \"Generate a wiki\""
fable-goal-cycle
synaptent
Run one bounded goal cycle - package live repo + agent-activity context, consult Claude Fable 5 for ranked next goals, a one-cycle plan, and exactly one paste-ready next prompt, then execute that prompt. Use at the top of each autonomous conductor cycle, when asked "what should the next goal/prompt/plan be", or when running a looped adaptive mission. The response is strategy input, not authority - tier gates, quorum, and the operating contract still decide what merges.
consult-fable
synaptent
Bounded advisory consult of Claude Fable 5 from inside the aragora repo. Use when you are stuck on prioritization ("what should I work on next?"), need a second opinion on a design or a reviewer deadlock, want a plan sanity-checked, or were asked to "ask Claude / Fable" something. Runs scripts/consult_claude.py with a per-attempt timeout and a derived total timeout, strict empty MCP config for CLI calls, and fail-closed backend attempts. Advice is input, not authority — it cannot approve merges, settle quorum, or override gates.
alexandrescu-modern-cpp-design
copyleftdev
Write C++ code following Andrei Alexandrescu's Modern C++ Design principles. Emphasizes policy-based design, template metaprogramming, and type-safe generic abstractions. Use when designing flexible, reusable libraries or when compile-time computation beats runtime overhead.
geo-content-optimizer
refly-ai
GEO内容优化工具，同时优化SEO和AI搜索引擎可见性，让内容更容易被ChatGPT/Claude等AI系统引用
agent-interop
docxology
Keep Codex, Claude Code, and Hermes workflows aligned through portable skills, explicit runtime adapters, and shared MCP contracts.
skill-creator
quran
Guide for creating effective skills. This skill should be used when users want to create a new
navigating-visual-logical-tree
christian289
Navigates WPF Visual Tree and Logical Tree with VisualTreeHelper and LogicalTreeHelper patterns. Use when traversing elements, accessing template internals, or understanding event routing.
solo-scaffold
fortunto2
Generate complete project from PRD + stack template — directory structure, configs, CLAUDE.md, git repo, and GitHub push. Use when user says "scaffold project", "create new project", "start new app", "bootstrap project", or "set up from PRD". Uses SoloGraph for patterns and Context7 for latest versions. Do NOT use for planning features (use /plan) or PRD generation (use /validate).
open-code-review
ToanMobile
Dùng khi cần review code tự động, kiểm tra chất lượng git diff trước khi commit/PR, hoặc audit toàn bộ file bằng OpenCodeReview CLI (Alibaba Group). Hỗ trợ cả chế độ OCR-managed và Delegation mode.
document-pdf
vasilyu1983
Extract text/tables from PDFs, create formatted PDFs, merge/split/rotate, handle forms and metadata. Supports pdf-lib/pdfkit (Node.js) and pypdf/pdfplumber/ReportLab (Python).
claude-code-commands
vasilyu1983
Create slash commands for Claude Code with $ARGUMENTS handling, agent invocation patterns, and template best practices. Reference for building user-triggered workflow shortcuts.
docs-codebase
vasilyu1983
Technical writing patterns for README files, API documentation, architecture decision records (ADRs), changelogs, contributing guides, code comments, and docs-as-code workflows. Covers documentation structure, style guides, Markdown best practices, and documentation testing.