This project addresses the challenge of knowledge loss and operational continuity when valuable team members depart. It enables users to create personalized AI "Skills" that emulate the technical expertise, communication style, and decision-making patterns of specific colleagues, thereby preserving institutional knowledge and facilitating seamless transitions. The target audience includes engineers, researchers, and power users who need to quickly onboard or maintain productivity despite personnel changes.

How It Works

The core approach involves generating a specialized AI "Skill" for a departed colleague using provided data sources such as chat logs, documents, and subjective descriptions. Each Skill comprises two integral components: a "Work Skill" encapsulating technical specifications and knowledge, and a "Persona" structured across five layers (rules, identity, expression, decision, behavior). This dual-component design allows the AI to not only execute tasks according to the colleague's technical proficiency but also to adopt their unique communication style and behavioral patterns, offering a deeply personalized emulation.

Quick Start & Requirements

Installation involves cloning the repository into specific directories for either Claude Code (.claude/skills/create-colleague or ~/.claude/skills/create-colleague) or OpenClaw (~/.openclaw/workspace/skills/create-colleague). Optional dependencies can be installed via pip3 install -r requirements.txt. Automated data collection from Feishu or DingTalk requires configuring App credentials, detailed in INSTALL.md. Links to official documentation include INSTALL.md and README_EN.md.

Highlighted Details

• Versatile Data Ingestion: Supports a wide array of data sources including Feishu (API/browser), DingTalk (browser), PDF, images/screenshots, email (.eml/.mbox), Markdown, and direct text pasting.
• Layered Persona & Work Emulation: Skills are built on a "Work Skill" for technical execution and a 5-layer "Persona" for nuanced behavioral and communication mimicry.
• Dynamic Evolution & Versioning: Skills can be continuously improved by merging new data, applying immediate dialogue corrections, and automatically archiving/rolling back to historical versions.
• Extensive Customization: Supports a rich set of personality tags (e.g., "甩锅高手" - master of shirking responsibility) and corporate culture styles (e.g., "字节范" - ByteDance style), along with specific company role/level mappings.

Maintenance & Community

The project is primarily maintained by titanwings. No explicit details regarding other contributors, sponsorships, or dedicated community channels (like Discord/Slack) are provided in the README, beyond the GitHub repository itself.

Licensing & Compatibility

The project is released under the MIT License, which is generally permissive for commercial use and integration into closed-source projects.

Limitations & Caveats

This project is currently in a beta testing phase. The quality and effectiveness of the generated AI Skills are directly contingent upon the quality and comprehensiveness of the input data provided. Automated collection for DingTalk is limited due to API restrictions on historical message retrieval.