Datasets:

hharsha
/

agentic-github-meta

license: mit
language:
  - en
task_categories:
  - text-classification
  - text-generation
tags:
  - agents
  - rag
  - llmops
  - mcp
  - multi-agent
  - github
  - tags
pretty_name: Agentic GitHub Meta
size_categories:
  - n<1K
dataset_info:
  features:
    - name: input
      dtype: string
    - name: target
      dtype: string
  splits:
    - name: train
      num_examples: 687
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/train-00000-of-00001.parquet

Agentic GitHub Meta

Text2text dataset of agentic AI / RAG / LLMOps / multi-agent GitHub-style descriptions paired with comma-separated tags. Schema mirrors zamal/github-meta-data (input, target) but the content focuses on projects from hharsha98 / hharsha plus synthetic paraphrases and search-style queries.

Column	Meaning
`input`	Repo description, search query, or paraphrase
`target`	Comma-separated tags (deduplicated, lowercased)

Train rows: 687

Sources

Rows expanded from hharsha/agentic-systems-showcase
Public GitHub repos under hharsha98 (descriptions + README paraphrases)
Synthetic queries such as “looking for multi-agent orchestration with MCP”, “RAG hybrid search recall@k”
Tags are honest topic labels (agents, multi-agent, rag, llmops, mcp, langgraph, fastapi, nextjs, observability, evals, docker, kubernetes, …) — not keyword spam

Intended use

Training small text2text models (e.g. T5 + LoRA) for GitHub-style tag generation over agentic/RAG projects.

Companion model: hharsha/agentic-github-tagger.

Studio: https://agentic-systems-studio.com · GitHub: https://github.com/hharsha98

License

MIT