consulting-writing
Management-consulting writing craft — McKinsey SCR (Situation·Complication·Resolution), Minto Pyramid/MECE, BCG bold-bullet executive summary, so-what upfront,…
Verifiable-attribution and citation-discipline craft — atomic claim decomposition, evidence grading (primary/analysis/forecast tiers), claimant attribution, citation typing (cites/references/contradicts/defines), source anchoring (Xanadu). Use when writing or reviewing a
$ npx -y skills add alfadur7/llm-wiki-newsroom --skill scholarly-citation --agent claude-codeHow it fires
How this skill gets triggered: by you, by Claude, or both.
/scholarly-citationContext preview
The summary Claude sees to decide when to auto-load this skill.
Verifiable-attribution and citation-discipline craft — atomic claim decomposition, evidence grading (primary/analysis/forecast tiers), claimant attribution, citation typing (cites/references/contradicts/defines), source anchoring (Xanadu). Use when writing or reviewing a
name: scholarly-citation description: Verifiable-attribution and citation-discipline craft — atomic claim decomposition, evidence grading (primary/analysis/forecast tiers), claimant attribution, citation typing (cites/references/contradicts/defines), source anchoring (Xanadu). Use when writing or reviewing a reference or synthesis page that attributes and cites source-based claims so they stay traceable, or when claim attribution, citation typing, or evidence grading is needed.
Verifiable-attribution craft drawn from scholarly/scientific citation traditions (citation classification, typing, anchoring). It ensures every claim in this wiki is atomized, evidence-graded, attributed to a speaker, typed by citation meaning, and anchored to a source location — and therefore traceable. Unlike the three borrowed traditions (narrative, exec summary, encyclopedic neutrality), this craft synthesizes external sources into an atomic-claim schema and is the project's signature. `criteria.json` is the SoT for each criterion's definition, comparator, and source; wiki-global state (page existence·hub type·section titles) is injected by the orchestrator (`tools/_lint/source.py`). The schema literals (`## Key Claims`, `[fact]`, `cites:` etc.) are the wiki's actual schema markers and section headers.
One claim, one line, one utterance. Each line of `## Key Claims` (Key Claims) marks an evidence grade — `[fact]` (primary source), `[analysis]` (secondary analysis), `[forecast]` (forecast). Putting two subjects' claims on one line (joined by "and"/"&" or a comma series spanning two hubs; ko: the conjunctions `와`/`및`), or chaining verb phrases on one line ("…did, and …did"; ko: `~했고, ~했다`), breaks atomicity.
Every claim is bound to who said it. The slot right after the grade marker names the speaker — a `[[wikilink]]` when they have a page, their plain-text real name and role when they fall below the page-creation threshold. What WP:ASF requires is the named source; the wikilink is one means to it, so demanding a link from a speaker who cannot have a page only pushes the author to link some other entity. Anonymous or collective subjects ("the government"·"industry sources"; ko `정부는`·`업계에서는`) stay forbidden in either form ("avoid stating opinions as facts"; attribute them to a named source), as does naming in plain text a page that does exist. A blockquote names its speaker (APA author-date principle — identify the cited author), wikilinked when that speaker has a page. A speaker with no page stays plain text with their role spelled out. Any target that is linked must exist as a real page.
Each line of `## Connections` (Connections) types the citation's meaning — `cites:` (direct grounds)·`references:` (context)·`contradicts:` (dispute)·`defines:` (definition). This extends scite Smart Citations' Supporting / Mentioning / Contrasting classes (with `defines:` added). If every line collapses to the default-safe `references:`, that signals avoidance of meaning classification. `defines:` points only to a concept hub.
An external-evidence citation anchors to the source's section — `[[slug#section]]` (ko: `[[slug#섹션]]`). This applies Xanadu citation anchoring (visible links to the precise source location, Ted Nelson) at the source level; a gradual migration, so 0% still passes (advisory), but the anchor target's slug and section must exist.
A piece synthesizing many sources follows Select→Read→Cite discipline — the declared source set (frontmatter) is the top set, and both the sources read and the sources cited in the body are subsets of it.
The synthesizing piece's narrative consciously reflects the evidence-grade (`[fact]`·`[analysis]`·`[forecast]`) and citation-type (`cites:`·`references:`·`contradicts:`·`defines:`) schema attached to sources — noting which grades the evidence base rests on, the dominant claimant, strong coupling (cites) vs context (references) — to deepen attribution. Confine attribution strength to factual statements; do not let it spill into verdict conclusions. Deterministic measurement (`count_grade_meta`·`count_cite_type_meta`) counts grade/cite-type meta expressions as advisory.
Each URL points to the relevant page as of the last verification.
A multi-agent AI knowledge base run by a five-role "newsroom" — open-source, local-first, no vendor lock-in.
Management-consulting writing craft — McKinsey SCR (Situation·Complication·Resolution), Minto Pyramid/MECE, BCG bold-bullet executive summary, so-what upfront,…
Encyclopedic neutral-reference writing craft — NPOV (attribute facts not opinions, due weight, neutral faction labels, verdict restraint), summary style and…
Guideline-authoring craft for instruction SoTs (.claude/ guides, CLAUDE.md, plan files) — operative rules vs recital, MUST/SHOULD/MAY force tiers, pruning,…
Journalism and argumentation writing craft — inverted pyramid, lede, nut graph, kicker, explainer framing, PAGE frames, Toulmin argument…
Forces the laziest solution that actually works, simplest, shortest, most minimal. Channels a senior dev who has seen everything: question whether the task…