Patriot University Glossary Editor
AI Skills

Patriot University Glossary Editor

Skip to main content
< All Topics
Print

Patriot University Glossary Editor

Posted
Updated
ByPU Publish
0 out of 5 stars
5 Stars 0%
4 Stars 0%
3 Stars 0%
2 Stars 0%
1 Stars 0%

Patriot University (patriot.university) runs a native Echo Knowledge Base glossary: a curated set of civic, legal, and constitutional concepts whose first occurrence in any KB article is auto-highlighted with a hover tooltip, and whose tooltip links to a document-network graph of every article that uses the term. This skill governs how editors research, write, and vet those entries. It composes with — and never overrides — patriot-editorial-framework (institutional voice, source-separation) and the patriot-private-citizen-inclusion-gate.

When to Use

  • Writing a new glossary definition, or editing an existing one.
  • Reviewing candidate terms produced by the extraction pipeline (pu_glossary.py extract) before publishing.
  • Deciding whether a phrase belongs in the glossary at all (inclusion/exclusion calls).
  • Deciding whether to suppress a term (block it from ever being proposed/published).

When NOT to Use

  • Writing article body prose — that is patriot-editorial-framework.
  • Building an accountability profile of a person — that is accountability-profile-builder. People are never glossary terms.

What the Glossary Is For

The glossary defines the reusable civic/legal vocabulary a skeptical, time-pressed, news-literate reader needs to verify and act on our reporting: doctrines (due process, prior restraint), statutes (Insurrection Act, Alien Enemies Act), procedures (cloture, class certification), and structural concepts (federalism, separation of powers). It is a reference layer, not commentary. A good entry lets a reader who hits an unfamiliar term get a neutral, grounded definition in one hover — then, if they want, see which other documents engage the concept.

Hard Constraints (enforced by the pipeline — do not fight them)

Constraint Rule
Definition length Plain text, ≤ 500 characters. No markdown, no HTML, no links, no quotation marks used for emphasis. EKB stores and truncates plain text.
Term name ≤ 100 characters; the natural form a reader sees in article text (lowercase common-noun concepts, Title Case named statutes/amendments).
One meaning EKB has no multi-meaning support. If a term is genuinely ambiguous, pick the single most civically relevant sense.
No stored per-doc links Highlighting is computed at render time. Publishing a term auto-highlights it everywhere; deleting it removes the highlights everywhere. There is nothing to wire per-article.

Definition Craft

  1. Concept-first, self-contained. The first sentence is a stable, neutral definition of the

concept that reads correctly out of context. Never open with “This document…”, a case caption, a date, or a person’s name.

  1. Grounded, not invented. Base the definition on how PU’s own knowledgebase uses the term

(RAG over the source docs), not free-form model knowledge. Never invent statutory citations, case names, or holdings — verify any citation appears in the KB.

  1. Attribute the contested part. When a definition includes applied or contested context

(how the concept has recently been used or challenged), attribute it — “Patriot University’s materials note…”, “courts have applied it to…” — rather than stating a contested claim as flat fact. Lead neutral; attribute the edge.

  1. Charged terms, neutral framing. Politically charged phrases (e.g. “big lie”, “enemy of the

people”) may be defined only as the rhetorical/political phenomenon they are, in neutral, educational language. If you cannot write it neutrally, defer it.

  1. Tone. Encyclopedic and civic-educational — the institutional CAF register, not opinion voice.

No editorializing, no adjectives that take sides.

Exclusion Guardrails (the inclusion gate for terms)

Reject a candidate — do not publish it — if it is any of these:

  • A person. Never. Individuals belong in accountability profiles, never the glossary. (This is

the patriot-private-citizen-inclusion-gate applied to terms: a person’s name auto-defined in a 500-char tooltip bypasses every editorial guardrail and is a defamation/tone hazard.)

  • An organization / named entity / proper noun. Agencies, offices, companies, PACs, named

commissions, outlets, tools, products, databases, or programs by name (ICE, the FCC, a specific bill or task force). The concept may qualify (“oversight”), the named body does not. Classic trap: “election integrity” is a concept and a fine term; “Election Integrity Network” is an organization and must be suppressed.

  • A one-off case, bill, or statute-by-citation that is not itself a durable, recognizable

concept. (Named foundational statutes that recur civically — Insurrection Act, NVRA — can qualify; a specific U.S.C. section or a single lawsuit does not.)

  • Generic or vague. “policy”, “process”, “reform”, “government”, “framework”, “system”,

“best practices” — not glossary-worthy.

  • A near-duplicate of a term already live or already selected this round (check the live list;

the pipeline dedups exact names but you must catch conceptual near-dups — e.g. “whistleblowing” vs “whistleblower”, three “subpoena” variants).

When in doubt on volume: a tight, high-signal glossary beats a sprawling one. Over-inclusion makes nearly every paragraph a highlighted term and destroys the feature’s value.

How an Entry Becomes Live (pipeline)

terms.json is the source of truth. The tooling lives in scripts/pu_glossary.py (+ make targets):

  1. Extract — pu_glossary.py extract --category X (or --slug, --docs, --new) scans docs

with Claude and proposes net-new candidates, deduped against the live glossary and the suppression blocklist, into a review file. It never auto-publishes.

  1. Curate (this skill) — apply the craft + exclusion rules above; keep the strongest concepts.
  2. Log & suppress — record dropped terms in curation-log.json; add true never-terms (orgs,

proper nouns) to suppression.json via pu_glossary.py suppress "".

  1. Merge & publish — pu_glossary.py merge into terms.json, then inject --apply (idempotent

upsert; auto-rebuilds and redeploys the document-network graph and purges cache).

  1. Remove — pu_glossary.py suppress "" --delete deletes a live term end-to-end (removes

from terms.json, deletes the live taxonomy term, rebuilds manifest + graph, purges) and blocks it from returning.

Editor’s Pre-Publish Checklist

  • [ ] Concept, doctrine, statute, or procedure — NOT a person or a named organization.
  • [ ] Definition ≤ 500 chars, plain text, concept-first, reads correctly out of context.
  • [ ] Grounded in the KB; any citation actually appears there; nothing invented.
  • [ ] Contested/applied context is attributed, not asserted as fact; neutral tone.
  • [ ] Not a near-duplicate of a live or just-selected term.
  • [ ] If borderline-charged, the framing is strictly neutral/educational — else defer.
Was this article helpful?
0 out of 5 stars
5 Stars 0%
4 Stars 0%
3 Stars 0%
2 Stars 0%
1 Stars 0%
5
Please Share Your Feedback
How Can We Improve This Article?
Table of Contents

Related Articles