HubLensLLMtitanwings/colleague-skill
// archived 2026-04-21
titanwings

colleague-skill

AI🌱 NEW PROJECT BOOST#LLM#Agent#Persona#Knowledge Distillation#Automation
View on GitHub
52

// summary

dot-skill is a versatile AI framework that distills individuals into interactive digital skills by analyzing their unique thought patterns and communication styles. The platform supports three distinct character families, including professional colleagues, personal relationships, and public figures. It integrates seamlessly with multiple AI agent hosts to provide a unified, automated experience for creating and invoking personalized AI personas.

// technical analysis

dot-skill is a sophisticated framework designed to distill human personas into AI-driven skills, evolving from a colleague-specific tool into a versatile engine for modeling colleagues, relationships, and public figures. Its architecture utilizes a two-layer system—Persona and Work/Context—to capture both the mannerisms and the decision-making frameworks of individuals. By supporting diverse data sources and providing a structured research toolchain, the project enables users to create highly accurate digital representations that think and speak in the target's frame, effectively solving the problem of preserving institutional knowledge or personal connection.

// key highlights

01
Supports three distinct character families: colleagues for professional workflows, relationships for personal connections, and celebrities for mental model reproduction.
02
Features a comprehensive six-dimension research toolchain for celebrities that processes subtitles, transcripts, and research notes into a coherent persona.
03
Implements an incremental learning system that allows users to provide feedback or corrections, which are then integrated into the persona without overwriting previous data.
04
Provides cross-host compatibility, allowing generated skills to run seamlessly across Claude Code, Hermes, OpenClaw, and Codex environments.
05
Includes automated data collection tools for platforms like Feishu, DingTalk, and Slack to streamline the ingestion of professional and personal communication history.
06
Maintains a robust version control system that archives every update, enabling users to easily roll back to previous iterations of a distilled skill.

// use cases

01
Distilling professional colleagues to replicate specific technical standards, workflows, and workplace communication styles.
02
Creating empathetic digital representations of personal relationships using chat history and emotional interaction patterns.
03
Developing celebrity or public figure personas through a six-dimension research toolchain that captures their mental models and decision frameworks.

// getting started

To begin, instruct your compatible AI agent (such as Claude Code or Hermes) to install the skill by providing the repository URL: 'Install the dot-skill skill for me: https://github.com/titanwings/colleague-skill'. Once installed, launch the interface by typing '/dot-skill' and follow the prompts to select a character family and provide the necessary source data to begin the distillation process.