tayyabexe/skills
Provide standard AI/ML task definitions as portable skills for use with major coding agents like OpenAI Codex and Claude Code.
What's novel
Provide standard AI/ML task definitions as portable skills for use with major coding agents like OpenAI Codex and Claude Code.
Code Analysis
7 files read · 2 roundsA collection of AI agent skill folders containing SKILL.md instruction files and Python helper scripts that guide AI coding tools through Hugging Face workflows including dataset creation, SQL-based dataset querying via DuckDB, model fine-tuning with TRL/Unsloth, evaluation metric extraction from mo
Strengths
The Python scripts contain genuine implementation depth — the SQL manager provides a full query engine over HF datasets with schema discovery, joins, and push-to-hub; the evaluation manager has sophisticated markdown table parsing with transposed table detection and model name normalization; the Unsloth training script is production-ready with proper LoRA config, eval splitting, and overfitting detection. The PEP 723 inline dependency approach makes scripts self-contained and portable.
Weaknesses
No test suite exists for any of the Python scripts despite complex logic like table parsing and SQL injection prevention; the README is written for a non-technical Windows audience (mentions antivirus, right-click extract) which misrepresents the project as a downloadable app rather than a developer/AI-tool resource; some scripts are essentially thin wrappers around huggingface_hub API calls with limited added value beyond CLI convenience.
Score Breakdown
Signal breakdown
Innovation
Craft
Traction
Scope
Evidence
Commits
168
Contributors
20
Files
104
Active weeks
13
Repository
Language
Python
Stars
1
Forks
0
License
Apache-2.0