61,650 agent skills collected from GitHub repositories, each containing a name, description, and full markdown content. The dataset was created by LittleDinoC and was last updated on May 23, 2026. It includes structured fields such as repository owner and path to the original markdown file.
Use Cases
- Training AI agents to understand and execute tasks based on skill descriptions and usage instructions.
- Building skill retrieval systems that match agent capabilities to task requirements using the skill name and description fields.
- Analyzing the structure and content of agent skills documented in markdown format across GitHub repositories.
- Creating benchmarks or evaluation datasets for agent performance on predefined skill sets.
Strengths
- Contains 61,650 distinct agent skills, providing a substantial corpus for analysis and training.
- Each entry includes a name, description, and the full markdown content, offering detailed context.
- Data is sourced from GitHub repositories, indicating a practical, real-world origin.
Limitations
- Row count is unknown, which may limit suitability assessment.
- Column-level documentation is absent; field semantics must be inferred after download.
- Description metadata is limited; actual data quality requires manual inspection after download.
Provenance
- Source
- GitHub repositories.
- Collection Method
- Collection from GitHub repositories, likely via scraping or API.
- Freshness
- Last updated 2026-05-23 14:38:52; freshness should be verified.