Stats
Actions
Tags
Help us improve
Share bugs, ideas, or general feedback.
From rampstack-skills
Guides humans and AI through content collaboration: participation boundaries, voice ownership, disclosure, and team calibration against AI slop. Triggers on AI writing workflows or generic-feeling AI content.
npx claudepluginhub rampstackco/claude-skills --plugin rampstack-skillsHow this skill is triggered — by the user, by Claude, or both
Slash command
/rampstack-skills:ai-content-collaborationThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
A senior editorial leader's playbook for how humans and AI compose in content workflows. Pragmatic, tool-agnostic, honest about both what AI in the loop enables and what it threatens.
references/ai-participation-boundaries.mdreferences/ai-slop-detection-and-avoidance.mdreferences/common-collaboration-failures.mdreferences/disclosure-and-transparency-patterns.mdreferences/ethics-and-intellectual-honesty.mdreferences/hybrid-workflow-patterns.mdreferences/quality-calibration-with-ai-in-loop.mdreferences/team-training-and-calibration.mdreferences/voice-ownership-preservation.mdPre-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.
Applies research-backed principles to craft human-like prose avoiding AI tells. For articles, blog posts, emails, marketing copy, social media—not code or docs.
Rewrites AI-generated text to sound natural by removing common AI tells like inflated significance and formulaic phrasing. Matches personal or brand voice from about/me.md or user samples.
Share bugs, ideas, or general feedback.
A senior editorial leader's playbook for how humans and AI compose in content workflows. Pragmatic, tool-agnostic, honest about both what AI in the loop enables and what it threatens.
Most content programs in 2026 use AI somewhere in the workflow. Pretending otherwise is dishonest; treating AI as a magic content factory is the failure mode this skill exists to prevent. The discipline is in between: knowing where AI legitimately accelerates, where humans must own, what hybrid patterns produce work that earns reader trust, and what crosses the line into AI slop or intellectual dishonesty.
This skill is the WORKFLOW layer that composes with every other content skill. Briefs can be AI-assisted; hub architectures can be AI-assisted; programmatic SEO is almost always AI-involved; editorial QA now includes AI-content audit by necessity. The collaboration discipline applies to all production stages, not to a single artifact type.
The voice is pragmatic and tool-agnostic deliberately. The methodology applies whether the AI in your loop is one of the major commercial models, an open-source model, or whatever ships next quarter. What stays constant is the workflow shape, the participation boundaries, the voice ownership question, and the ethical frame. What changes is which specific tool you reach for, which is implementation work that varies by team and budget.
When to use this skill: building or refining an AI-content workflow, calibrating a team on consistent AI usage, addressing the "we use AI but our work feels generic" problem, designing disclosure policies, or working through the ethics of AI-assisted content production for a regulated or trust-sensitive context.
This skill spans the workflow layer of AI-assisted content production. It composes with all six other content-suite skills as the cross-cutting discipline.
content-strategy is program scope: what to produce. Strategy decisions can be AI-assisted; the program-level judgment stays human.pillar-content-architecture is hub scope: how the topical hub fits together. Hub architecture can be AI-suggested; the architectural commitment stays human.content-brief-authoring is per-piece scope: briefs each piece. Briefs can be AI-drafted from research; the contract decisions stay human.content-and-copy is execution scope: writes each piece. Drafts can be AI-produced; voice and editorial judgment stay human.programmatic-seo is scaled scope: generates pages from data. AI generation is the dominant production model; sampling QA is the human gate.editorial-qa is gate scope: verifies before publish. AI-content audit is now a load-bearing gate; the audit's judgment stays human.The audience: editorial leaders, content directors, content ops managers, agencies running AI-assisted production, in-house teams calibrating AI usage across writers. The voice is senior editorial leader to junior editor or content marketer. Pragmatic, honest, tool-agnostic.
What is not in scope: specific prompts (those are implementation; teams develop their own), specific tool endorsements (the methodology applies regardless of which tool is in the loop), specific integration code (varies by stack and team). Tool categories appear when they earn methodology relevance; specific tools appear only as illustrations of categories, never as recommendations.
The keystone framing.
The pathology to avoid is treating AI as either a magic content factory (cheap, fast, scaled, output quality optional) OR as a forbidden intruder (purity gospel that does not survive contact with deadlines). Both readings produce bad work.
The discipline that produces durable work: humans own the content; AI accelerates the work. Specifically:
Humans own. Editorial judgment, voice, distinctive POV, fact accuracy, ethical decisions, what to publish versus what to kill, brand voice, narrative arc, tone calibration, reader empathy, claim verification.
AI accelerates. Research synthesis, draft generation against a brief, copy edit suggestions, alternative phrasings, summary, transcription, quality-control automation at scale.
The line. AI does work that the human directs and verifies. AI does NOT make decisions about what publishes, who is quoted, what is true, or what voice the brand uses.
The litmus test. If your AI-assisted piece publishes without a human being able to defend every claim, every position, and every word, you have crossed the line. The piece is AI's work, dressed in your byline. Readers eventually notice.
A non-exhaustive list of stages where AI in the loop is fine and often improves the work.
In each case, AI accelerates work the human still owns. The acceleration is real; the ownership stays unchanged.
Detail in references/ai-participation-boundaries.md.
The boundary list.
The "human in the loop" framing is necessary but insufficient. A human briefly reviewing AI-generated content before publish is not ownership; it is rubber-stamping. Ownership requires the human to have made the actual decisions the piece embodies.
Five patterns that work, with tradeoffs.
1. AI-first draft, human-edit-heavy. AI produces a 90% draft; the human spends 60% of the time editing. Output: efficient for high-volume editorial; risks generic voice if editing is light.
2. Human-first outline + research, AI-draft, human-rewrite. Human builds the outline and gathers research; AI drafts within that scaffold; human rewrites in voice. Output: preserves voice better; slower than AI-first.
3. AI-as-research-assistant, human-writes. AI condenses sources into a brief; human writes the entire piece from the brief. Output: highest voice fidelity; slowest.
4. Human-writes, AI-as-editor. Human drafts; AI suggests edits, alternative phrasings, copy edits; human accepts or rejects. Output: writer voice preserved; AI catches details.
5. AI-generates-at-scale, human-samples. For programmatic SEO. AI generates thousands of pages; human samples 50 to 200 with editorial-qa discipline. Output: scaled production; depends entirely on template quality and sampling discipline.
The pattern that fits depends on volume, voice sensitivity, team skill, and time budget. No pattern is "the right one"; pattern selection is a real decision that should match the production context.
Detail in references/hybrid-workflow-patterns.md.
Voice is the dominant casualty of careless AI workflows. The patterns that preserve voice.
The honest framing. Voice is the hardest thing to preserve in AI-assisted work and the easiest thing to lose. Programs that do not actively preserve voice end up with content that is technically correct, semantically generic, and indistinguishable from competitors using the same tools.
Detail in references/voice-ownership-preservation.md.
AI slop is the term of art for AI-generated content that is technically functional but reads as generic, derivative, and signal-less. Cross-reference editorial-qa's ai-content-audit-patterns reference for the detection patterns; this section addresses prevention.
Patterns that produce slop.
Patterns that prevent slop.
content-brief-authoring)The reader-detection problem. Readers can often sense AI-flavored content even when they cannot articulate why. Generic openings, predictable structures, "perfect" grammar that is emotionally flat. Slop loses reader trust over time even when individual pieces are not penalized.
Detail in references/ai-slop-detection-and-avoidance.md and cross-reference editorial-qa's audit patterns.
When should AI usage be disclosed to readers?
The tiered framework.
The principle. Disclose when the reader's understanding of the content's origin would change their trust in it. A bylined opinion piece purportedly by a named expert that is substantially AI-drafted is a trust violation; a product description on an ecommerce site that was AI-drafted is not.
Disclosure language patterns (when used).
Industry-specific norms vary. Major journalism organizations have published explicit AI usage standards. Content marketing has weaker norms but is moving toward disclosure for high-trust pieces.
Detail in references/disclosure-and-transparency-patterns.md.
Inconsistent AI usage across a team produces inconsistent output. The discipline.
The pathology. AI usage emerges informally, every writer develops their own patterns, output drifts, editors cannot pinpoint why pieces feel off. The discipline is making AI usage explicit, calibrated, and documented.
Detail in references/team-training-and-calibration.md.
AI tools were trained on copyrighted material. That is the simple ethical reality of every major LLM in 2026. The catalog's position on this question is not "AI use is unethical" (that would render the catalog itself hypocritical) but "intellectual honesty about AI involvement is non-negotiable."
The principles.
The intellectual-honesty frame supersedes any specific policy debate. Teams that treat AI usage with intellectual honesty produce content readers can trust over time. Teams that hide, deny, or rationalize lose trust eventually.
Detail in references/ethics-and-intellectual-honesty.md.
Rapid-fire. Diagnoses in references/common-collaboration-failures.md.
When designing or auditing an AI-assisted content workflow, walk these 12 considerations.
The output of the framework is a workflow document the team can reference: AI participation rules named, hybrid pattern selected, voice preservation patterns specified, disclosure tier set, calibration cadence committed, ethical floor articulated, accountable signer named for each piece.
references/ai-participation-boundaries.md - Where AI legitimately helps, where humans must own. The boundary list and the "human-in-the-loop is not ownership" distinction.references/hybrid-workflow-patterns.md - Five workflow patterns with tradeoffs and selection criteria. When each pattern fits production context.references/voice-ownership-preservation.md - Voice guidelines as prompt input, sample text as voice anchor, mid-draft voice check, final pass in human voice, reject-the-bland discipline.references/ai-slop-detection-and-avoidance.md - What produces slop, what prevents it. Cross-references editorial-qa's audit patterns.references/disclosure-and-transparency-patterns.md - Tiered disclosure framework, language patterns, industry norms.references/team-training-and-calibration.md - Documented policy, calibration sessions, voice library, quality benchmarks, onboarding.references/quality-calibration-with-ai-in-loop.md - How editorial standards shift when AI is in the workflow. Same standards, different failure modes.references/ethics-and-intellectual-honesty.md - Training data, attribution, fabrication boundaries, intellectual honesty as the supervening frame.references/common-collaboration-failures.md - 11+ failure patterns with diagnoses and fixes.AI in content workflows is neither magic nor menace. It is a category of tooling that, like every tooling category before it, rewards disciplined use and punishes careless use. The teams producing memorable AI-assisted content are the ones holding the line on human ownership, voice, fact accuracy, and intellectual honesty. The teams producing AI slop are the ones treating AI as a content factory.
The discipline is not anti-AI; it is pro-craft. Craft was always what made content worth reading; AI does not change that, it just raises the cost of skipping it.
When in doubt about whether an AI-assisted workflow is ready, ask: is human ownership specified, are participation boundaries documented, is voice preservation built into the prompt and review patterns, is fact verification a halt-condition, is disclosure tiered to audience trust, is the team calibrated, and is the ethical floor explicit? If yes to all of those, the workflow is ready. If no to any, the gap is where the program will produce slop and lose reader trust.