A hands-on tutorial that teaches AI agent building from scratch with local LLMs via node-llama-cpp, covering function calling, memory and ReAct.
GeminiLight/SocialCoach
An AI social-skills coach that lets you rehearse difficult conversations like raises, boundaries and conflicts through role-play, then gives feedback based on what you said.
- Estrellas
- 272
- Forks
- 44
- Issues abiertos
- 0
- Último commit
- hace 7 días
¿Qué es?
SocialCoach is a web app for practicing everyday social conversations, such as asking your boss for a raise, declining extra work, or asking a friend to repay a loan. You pick a scenario from a library of 58 across seven life situations, or describe your own situation, and then talk with an AI character who has its own goals and will push back, ask follow-up questions and hold its position.
After each session, the coach debriefs by quoting your own words before commenting, and separates how well you communicated from whether the conversation reached the outcome you wanted. Practice sessions are recommended based on your goals and estimated skill levels across 34 social-emotional skills, and the coaching draws on a sourced library of strategies and cases. The interface and content are available in Chinese and English, and the app is designed to run without an account.
The project is built with Next.js, React, Three.js and TypeScript, and works with Anthropic or OpenAI-compatible model providers. Users can bring their own API keys, and the repository includes Docker Compose and Vercel deployment setups.
Funciones clave
- 01Role-play conversations where the AI character responds from its own position and goals.
- 02Debriefs that quote your own words and separate communication quality from the outcome.
- 03Personalized practice recommendations based on goals and skill estimates across 34 skills.
- 04A library of 58 scenarios across seven life situations, in Chinese and English.
- 05Bring-your-own model keys for Anthropic or OpenAI-compatible providers, with local data export.
- 06Optional timed responses for practicing under pressure.
Casos de uso
- Rehearsing a salary negotiation or performance conversation before it happens.
- Practicing how to set boundaries with friends, roommates or family.
- Preparing for a tense meeting or a conflict with a coworker.
- Building social-emotional skills for personal growth in a private, account-free setting.
Temas
Preguntas frecuentes
- ¿Qué es SocialCoach?
- An AI social-skills coach that lets you rehearse difficult conversations like raises, boundaries and conflicts through role-play, then gives feedback based on what you said. SocialCoach is a web app for practicing everyday social conversations, such as asking your boss for a raise, declining extra work, or asking a friend to repay a loan. You pick a scenario from a library of 58 across seven life situations, or describe your own situation, and then talk with an AI character who has its own goals and will push back, ask follow-up questions and hold its position.
- ¿SocialCoach es gratuito? ¿Qué licencia tiene?
- SocialCoach es open source bajo la licencia Apache-2.0. Revisa los términos de la licencia antes de usarlo en productos comerciales.
- ¿SocialCoach sigue mantenido?
- Mantenimiento activo. El último commit de SocialCoach fue el 12 oct 2026.
- ¿Para qué puedo usar SocialCoach?
- Rehearsing a salary negotiation or performance conversation before it happens. Practicing how to set boundaries with friends, roommates or family. Preparing for a tense meeting or a conflict with a coworker. Building social-emotional skills for personal growth in a private, account-free setting.
- ¿Dónde consigo SocialCoach?
- El código fuente y las instrucciones de instalación están en GitHub: https://github.com/GeminiLight/SocialCoach
Resumen generado por IA a partir del README del proyecto. Las estadísticas de GitHub se actualizan a diario.
Proyectos relacionados
A step-by-step educational project that builds Retrieval-Augmented Generation from scratch in JavaScript using local LLMs and no cloud APIs.
Companion code for the book Designing Multi-Agent Systems, including PicoAgents, a multi-agent framework built from scratch for teaching.
A curated awesome-list of platforms, engines, UIs, models, tools and tutorials for running LLMs locally on your own hardware.
A guide to running autonomous AI coding loops with Geoff Huntley's Ralph methodology, covering planning, building, specs and prompt templates.
Motor de IA autohospedado y compatible con OpenAI que ejecuta LLM, modelos de visión, voz, imagen y video en cualquier hardware, incluso solo con CPU.