Stats
Actions
Tags
Help us improve
Share bugs, ideas, or general feedback.
From rampstack-skills
Pre-publish QA framework covering brief adherence, voice consistency, fact accuracy, AI-content audit, SEO/AEO compliance, and sampling at scale for editorial, AI-generated, and programmatic content.
npx claudepluginhub rampstackco/claude-skills --plugin rampstack-skillsHow this skill is triggered — by the user, by Claude, or both
Slash command
/rampstack-skills:editorial-qaThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
A senior editor's playbook for pre-publish content QA. The discipline that catches problems BEFORE content ships, not after.
references/ai-content-audit-patterns.mdreferences/brief-adherence-checklist.mdreferences/common-qa-failures.mdreferences/fact-accuracy-and-citation-discipline.mdreferences/internal-linking-and-schema-validation.mdreferences/qa-at-scale-patterns.mdreferences/qa-workflow-templates.mdreferences/seo-aeo-compliance-checklist.mdreferences/structure-and-clarity-review.mdreferences/voice-consistency-patterns.mdRuns unified pre-publish quality gate on marketing content: hallucination detection, claim verification, brand voice scoring, structure validation. Use before publishing copy.
Scores existing content against E-E-A-T and CITE rubrics for AI search readiness. Analyzes word count, citations, H-tag hierarchy, and page type to produce a publish/fix/no-publish verdict with veto items and top fixes.
Audits content quality against 80 CORE-EEAT criteria across 8 dimensions, producing per-item scoring, veto checks, and a prioritized fix plan. Useful before publishing or when evaluating E-E-A-T readiness.
Share bugs, ideas, or general feedback.
A senior editor's playbook for pre-publish content QA. The discipline that catches problems BEFORE content ships, not after.
Most content QA is broken in one of two directions. The thin version is "did I read it once and is the spelling OK," which catches typos but misses brief drift, voice inconsistency, hallucinated facts, AI tells, and structural problems that reach readers as "this is fine but not memorable." The thick version is a 47-item checklist that nobody completes honestly because it is process theater: checkboxes nobody actually believes catch problems.
This skill is the discipline of catch-problems QA. Each check earns its keep by catching a specific class of failure that would reach readers if missed. Checks that do not catch anything get cut. Checks the production volume cannot sustain get redesigned (sampling instead of full audit, automated instead of manual). The QA framework is what is left when you remove the theater.
The skill covers three production shapes: single editorial pieces (one at a time, full QA), AI-generated drafts (with the AI-content audit that did not exist as prominently 2 years ago), and programmatic SEO sets at scale (sampling discipline, threshold gating). Each needs its own QA shape; the underlying methodology composes across all three.
When to use this skill: building a content QA process from scratch, auditing an existing QA process that ships sloppy work or burns out the team, designing QA gates for an AI-assisted workflow, or building sampling discipline for programmatic SEO sets.
This skill spans pre-publish quality control. It plugs in at the END of every other content skill's output. The six-skill content suite distinction:
content-strategy is program scope: what to produce.pillar-content-architecture is hub scope: how the topical hub fits together.content-brief-authoring is per-piece scope: briefs each piece.content-and-copy is execution scope: writes each piece.programmatic-seo is scaled scope: generates many pages from data.Every skill above produces a draft. This skill is what gets drafts to publishable. It is the gate where quality is actually enforced.
The audience: editorial leads, content directors, in-house content QA, agencies with production lines, content ops managers, anyone running a writer (human or AI) and accountable for what ships. The voice is senior editor to junior editor or content marketer. Specific, opinionated, honest about where QA earns its keep versus where it becomes process theater.
The keystone distinction. Two failure modes plus the discipline.
Thin QA (typo-checking dressed as quality control). "I read it once, it is fine." Catches obvious mistakes; misses brief drift, voice inconsistency, hallucinated facts, structural problems, AI tells. Output: shipped content that is "fine but not memorable." Cost: invisible until a brand misstep, a hallucinated statistic, or a competitor's content compounds and the thin set falls behind.
Thick QA (47-item checklist nobody completes honestly). Every conceivable check listed; no triage. Reviewers either skim and check boxes (theater) or burn out under the cognitive load. Output: shipped content slightly more polished than thin QA produces, but the team's review velocity collapses. Cost: throughput drops; reviewers leave; the checklist atrophies into a few checks that actually run.
Catch-problems QA (the discipline). Each check earns its keep by catching a specific class of failure that would reach readers if missed. Checks that do not catch anything get cut. Production volume drives sampling versus full-audit decisions. Reviewers are accountable for what they caught, not for box-completion.
The litmus test. If the team can name the last 3 problems each QA check caught, the check earns its keep. If a check has not caught anything in 6 months, it is theater. Cut it; reallocate the attention to checks that catch real problems.
Did the writer execute the brief? The check is straightforward when the brief is well-authored (see content-brief-authoring).
The brief-adherence checks:
The brief-adherence check is the cheapest, fastest, highest-value QA gate. It runs first in the QA sequence because catching a brief-adherence failure early saves the editor from spending time on voice and structure on a piece that will need to restart.
If briefs are vague, this check is impossible. Fix briefs first; the QA process cannot enforce a contract that does not exist.
Detail in references/brief-adherence-checklist.md.
Does the piece sound like the brand?
For AI-co-authored pieces, voice drift is the dominant failure mode. AI assistants regress to a model-default voice unless the writer actively pulls them back. The QA check needs to read for the brand voice as much as for the words.
Detail in references/voice-consistency-patterns.md.
Every claim in the piece needs to be true. The check:
Hallucination is the dominant failure mode in AI-assisted writing. AI assistants generate plausible statistics, plausible quotes, plausible case studies, none of which are real. The fact-accuracy check is the gate that catches them. If you skip this gate on AI-generated content, you ship hallucinations.
Citation discipline:
Detail in references/fact-accuracy-and-citation-discipline.md.
Does the piece work as a piece?
Detail in references/structure-and-clarity-review.md.
The QA check that did not exist as prominently 2 years ago. AI-co-authored content has detectable patterns even when written by a competent human editor.
AI tells (pattern recognition).
Hallucination patterns (factual errors).
Voice drift.
The audit shape. Read with these patterns top-of-mind. Flag any match. The bar is not "AI was used" (it almost always is now); the bar is "would a careful human editor have shipped this." If the patterns above are present, the piece needs another revision pass with the patterns called out.
Detail in references/ai-content-audit-patterns.md.
Does the piece serve search engines AND answer engines?
SEO checks.
AEO checks.
Detail in references/seo-aeo-compliance-checklist.md.
Internal linking.
Schema validation.
Detail in references/internal-linking-and-schema-validation.md.
For pieces shipping at programmatic-SEO scale (100s to 100,000s of pages), full-audit QA is infeasible. Sampling discipline replaces it.
Sampling strategy.
Automated checks at scale.
Manual checks on sampled pages.
Threshold gating.
Detail in references/qa-at-scale-patterns.md.
Ownership and sequencing.
Single QA owner per piece, not a committee. The owner is accountable for what shipped. Committees diffuse accountability; nobody owns a problem that reaches readers.
Sequencing. Brief-adherence, then fact-accuracy, then structure, then AI-content audit, then voice, then SEO/AEO, then internal linking, then schema. Brief-adherence first because it is the cheapest gate; SEO/AEO checks last because they are easiest to fix and rarely halt-conditions.
Halt vs flag vs auto-fix.
Escalation. When QA finds a pattern (multiple pieces failing the same check), escalate to the brief author, the writer, or the editorial process owner. Pattern signals process problem, not just per-piece problem.
Detail in references/qa-workflow-templates.md.
Rapid-fire. Diagnoses in references/common-qa-failures.md.
When designing or auditing a QA process, walk these 12 considerations.
The output of the framework is a QA process the team can run repeatably: each check named, each owner named, each halt-condition documented, each sampling rule specified for programmatic surfaces.
references/brief-adherence-checklist.md - Every brief field as a QA check, with how to verify and what failure looks like.references/voice-consistency-patterns.md - Vocabulary, rhythm, stance, register, mid-piece sampling discipline for long pieces.references/fact-accuracy-and-citation-discipline.md - Verification methodology, hallucination detection, citation rules, source-age guidelines.references/structure-and-clarity-review.md - Lede patterns, sectioning principles, endings, structural anti-patterns.references/ai-content-audit-patterns.md - 11 AI tells, 6 hallucination patterns, voice drift detection, worked example with revision.references/seo-aeo-compliance-checklist.md - SEO and AEO checks combined into one workflow.references/internal-linking-and-schema-validation.md - Link discipline, anchor text variation, schema patterns and validation.references/qa-at-scale-patterns.md - Sampling strategy, automated checks, manual review at scale, threshold gating.references/qa-workflow-templates.md - Ownership, sequencing, halt versus flag versus auto-fix, escalation patterns.references/common-qa-failures.md - 11+ failure patterns with diagnoses and fixes.Every other skill in the content suite produces drafts. QA is the discipline that turns drafts into publishable work. It is also where every previous decision (brief shape, voice guidelines, hub architecture, programmatic template) gets tested against actual output. Skipping QA is not "moving fast"; it is shipping the failure modes of every upstream decision unfiltered.
The QA process is the immune system of the content program: invisible when it is working, catastrophic in its absence.
When in doubt about whether a QA process is ready, ask: does each check name a class of failure it catches, has each check actually caught something in the last 6 months, is brief-adherence first in the sequence, is fact-accuracy a halt-condition, is the AI-content audit included, does sampling discipline apply to scaled surfaces? If yes to all of those, the process is real. If no to any, the gap is where readers will encounter the failure that QA did not catch.