MCP App StoreMCP App Store
de crush-skills180
Créez et faites évoluer une compétence conversationnelle personnalisée de "crush" à partir de journaux de chat, photos et réseaux sociaux ; génère une Mémoire Relationnelle et un Persona pour la simulation.
Cette compétence aide les utilisateurs à distiller les interactions publiques/privées d'une personne réelle en une compétence "crush" locale et hors ligne. Elle guide l'entrée des données (alias, informations de base, personnalité), ingère le matériel source (exports de chat, captures d'écran, photos) et génère deux artefacts centraux : une Mémoire Relationnelle et un Persona. La compétence peut écrire des fichiers versionnés, prévisualiser des résumés et évoluer à mesure que de nouveaux éléments sont ajoutés. Elle impose des limites de sécurité pour éviter le harcèlement ou les violations de la vie privée.
Utilisez cette compétence lorsqu'un utilisateur souhaite un simulateur conversationnel privé pour une réflexion personnelle, une reconstruction de souvenirs ou des scénarios de jeu de rôle légers. Elle est destinée à l'analyse émotionnelle personnelle, et non pour contacter ou harceler des personnes réelles. Activez-la lorsque l'utilisateur demande de créer, mettre à jour ou lister des compétences crush.
Conçue pour les agents pouvant utiliser des outils Read/Write/Edit/Bash et gérer des flux de travail basés sur des fichiers (style Claude Code ou environnements d'agents similaires).
A Claude Code skill for creating simulated persona Skills from chat logs, photos, and social media about a crush. The SKILL.md is extensive and bilingual (Chinese/English) with detailed multi-step flows for intake, analysis, and persona generation. No scripts were bundled for execution. The skill references several Python tools (wechat_parser, qq_parser, social_parser, photo_analyzer, version_manager, skill_writer) via ${CLAUDE_SKILL_DIR}/tools/ but none were included for audit. Safety boundaries are explicitly stated (no stalking, local-only data, Layer 0 rules preventing out-of-character generation). However, the core premise of distilling a real person into a simulable persona raises privacy and ethical concerns, even with stated guardrails. The `rm -rf crushes/{slug}` delete command is present but scoped to the project directory.
Ethically ambiguous skill — simulates real people from their private data (chat logs, photos, social media). Has explicit safety boundaries but the guardrails rely on the LLM honoring Layer 0 rules with no enforcement mechanism. The privacy claim (local-only) is unverifiable. No scripts bundled so the skill is incomplete and cannot function as-is. The skill is well-structured for its niche but serves a narrow audience (Chinese-speaking users wanting to simulate crush conversations). Shell injection risk is low — variable interpolation in bash commands uses ${CLAUDE_SKILL_DIR} which is an env var, and {slug} comes from user input but is used in mkdir/write paths, not directly in shell commands with unquoted expansion.
Bounty Hunter — Atlas
Persona skill: 'Atlas' — a profit-focused developer persona for discovering, evaluating and executing paid bounties or freelance tasks with ROI-aware workflows.
Caveman Compress
Compress natural-language memory files into a compact 'caveman' format to save LLM tokens while preserving code, links, and structure.
Cost Tracker
Monitor agent session costs, set budget alerts, and get actionable token-spend optimizations to keep multi-session workflows within budget.
Chinese Software Copyright Application
Generate complete Chinese software copyright application materials (application form, code documentation, user manual, design docs) and export them to PDF from
Audit Agents/Skills/Commands
Run a quantitative quality audit on agents, skills, and commands to check production readiness and get prioritized recommendations.
Go Data Structures
Authoritative guidance on choosing and using Go built-in and standard-library data structures, with practical best practices for slices, maps, arrays, container
NotebookLM Research Agent
End-to-end research-to-content pipeline: ingest sources into Google NotebookLM, run deep cited research, generate downloadable artifacts, and produce polished c
Memory Search
Search past coding sessions and observations using natural-language queries to retrieve decisions, notes, and context.