academic-terminology-glossary

Scan the document for domain-specific terms, insert anchor links at each occurrence, and append a glossary table at the end. This Skill automates all three steps using a four-column format covering definition, context, project relevance, and references.

Define an anchor with `<a id="term-name"></a>` in the glossary table, then link to it in the body text using `[keyword](#term-name)` or a tagged code like `[[TN1]](#term-name)`. The Skill enforces exact matching between anchors and links.

It covers four groups: traditional security analogies like Prepared Statements and NX-bit, Transformer mechanisms like Self-Attention and Flat Token Space, alignment theory like Competing Objectives and Refusal Boundary, and PI-Guard methods like Group-Aware Splitting and INT8 quantization.

Run `python Final-Report/scripts/validate_local.py` after inserting terms. The script checks Markdown syntax, and you should also verify that every `<a id="term-..."></a>` anchor matches its corresponding `#term-...` link exactly.

Detection relies on a predefined keyword taxonomy, so novel or project-specific terms outside the four listed groups may be missed. Definitions and citations still require human review to ensure academic accuracy before final submission.