tiantian180/skills: A Collection of Practical AI Agent Skill Packs

tiantian180/skills is a collection of practical AI Agent skill packs, designed to help AI assistants acquire professional capabilities in specific domains. The project includes two core skills: glossy-reader (an interactive technical article intensive reading tool) and video-subtitle-translate (an end-to-end video subtitle translation pipeline). Adopting a modular design, each skill is an independent capability unit that supports reuse, expansion, and combination, and can be integrated into mainstream AI Agent frameworks.

In the AI Agent ecosystem, general large model conversations struggle to meet the needs of complex tasks in specific domains. Skill is an emerging concept in the AI Agent field; it is not just a prompt or script, but a complete capability package that includes workflows, templates, domain knowledge, and tool scripts. The tiantian180/skills project uses a modular design—each Skill has an independent directory, configuration files, and documentation, making it easy to reuse, share, and customize for expansion.

The glossy-reader skill focuses on enhancing the reading experience of technical articles:

1. Term Dictionary Bubble: Automatically identifies professional terms; click to expand explanations while maintaining reading continuity;
2. Paragraph Bilingual Switch: Supports original text/translation comparison, with a draggable floating window to adjust layout;
3. Plain Language Explanation Bar: The right sidebar simplifies technical content using analogies and daily examples;
4. Detail Optimizations: Image cache and zoom, day/night themes, safe link preview. Technically, it outputs an independent HTML page, supports multiple language pairs, and is suitable for blogs, documents, or Agent article processing workflows.

The video-subtitle-translate skill implements an end-to-end video subtitle translation pipeline:

1. Video Download: Supports multiple platforms (YouTube, Bilibili, etc.) based on yt-dlp;
2. Subtitle Extraction/Transcription: Prioritizes extracting embedded subtitles; if none exist, uses AI for transcription;
3. Translation: Can connect to large models or professional APIs;
4. Subtitle Burning: Generates bilingual/monolingual subtitle videos. Application scenarios include content creators making bilingual videos, learners converting foreign language tutorials, researchers processing materials in batches, and it can also serve as a pre-step for Agent video understanding.

Project design philosophy:

• Practicality First: Each Skill solves specific real-world problems (e.g., improving reading and translation efficiency);
• Experience First: Focuses on polishing details (such as the reading experience of glossy-reader and automation of translation processes);
• Modular and Reusable: Independent Skill design for easy combination and expansion. Contributing new Skills requires following: clear README, runnable examples, modular code, and reasonable default configurations.

tiantian180/skills can be integrated into Agent frameworks like OpenClaw, LangChain, AutoGPT, etc., and called as Tools. For example, a research assistant Agent can load both skills to process articles and videos, then combine with large models for analysis and summary. Integration considerations:

1. Provide clear input/output interfaces;
2. Handle errors and boundary conditions;
3. Document the scope of capabilities and limitations.

Currently, the project includes 2 highly completed Skills and is open-sourced under the MIT license. Future outlook:

• Expand Skills to more domains (data processing, code analysis, office automation, etc.);
• After enriching the Skill library, building domain-specific AI Agents will be as simple as building blocks;
• The project focuses on practicality and composability, has long-term value, and is worth developers' attention and participation.