File classifier
[COLM'26] SkillLearnBench is the first benchmark for evaluating continual learning methods that automatically generate agent skills.
npx -y skills add cxcscmu/SkillLearnBench --skill file-classifierAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
What its author says it does
Copied from the file, not written here
Use this skill to classify PDF, PPTX, or DOCX files into subject categories: LLM, trapped_ion_and_qc, black_hole, DNA, or music_history. Analyze filenames and available content to determine the correct category.
SKILL.md
1.4 KB, as published. Nobody here has run it
File Classifier Skill
Goal
Classify a document into exactly one of five subject folders:
- LLM
- trapped_ion_and_qc
- black_hole
- DNA
- music_history
Classification logic
- Analyze File Metadata (Name): Often, the filename contains keywords.
- Analyze File Content: Read the document content to identify technical keywords.
- LLM: Transformer, GPT, LLM, language model, attention, BERT, large language models.
- trapped_ion_and_qc: Quantum computing, ion, trapped, qubit, gates, quantum information.
- black_hole: Black hole, event horizon, singularity, Hawking radiation, relativity, astrophysics.
- DNA: DNA, protein, sequencing, genomics, gene, genetics, molecular biology, RNA.
- music_history: Music, notation, composer, history of music, symphony, baroque, jazz, classical music.
- Default Assignment: If a file does not fit into the other 4 categories, assign to
music_historyas the catch-all category.
Execution
For each file in the root:
- Use
read_fileorpdftool to sample content. - Apply the classification logic.
- Move the file to the determined target folder.