Researched Article Writer with Traced Citations and Your Voice
Traces each fact to its original source in a claim ledger, then drafts an outline or article in a voice measured from your posts.
Before & after
Same prompt, same agent — the only difference is the skill.
I can write a blog post about remote team productivity for you. To make it sound like your voice, it would help if you could share some examples of your previous writing or your brand voice guidelines. I could do some research on the topic and draft an outline first, then write the full post. You'd probably want citations or references so it's credible. I'd need to know your target audience and what angle you want to take—productivity tips, challenges, case studies, or something else. Want me to start with an outline?
VOICE ANALYSIS: Your Existing Content Analyzed 6 prior posts; voice profile: - Tone: Direct, skeptical of hype, occasional profanity (mild) - Sentence structure: Short-medium, active voice dominates - Proof method: Concrete metrics over theory; personal experience valued - Audience comfort: B2B ops/tech, prefers nuance over absolutes - Emoji use: None (professional restraint) - Depth: ~1,500-2,000 words; includes a contrarian take RESEARCH FINDINGS Sources queried: 12 remote-work studies, 3 productivity platforms, 5 case studies Cited sources: McKinsey remote work ROI (2024), Slack Workforce Index (2025), Buffer Compensation Study DRAFT OUTLINE (your voice) Introduction: "Remote productivity is broken because we measure the wrong thing" - Subvert expectation (most posts say "remote works"; you say "we're measuring wrong") - Teaser: Synchronous work metrics fail async teams Section 1: "The Attendance Trap" (350 words, data-driven) - Metric: 62% of managers still track hours-online despite evidence it correlates 0.18 with output - Contrarian: Visible-hours culture produces theater, not work - Cite McKinsey data: 40% of remote workers report they're visible-work maxing (low-impact tasks) Section 2: "What Actually Works" (400 words, based on 3 case studies) - Output-based OKRs (not activity logs) - Case study: SaaS company dropped stand-ups, output stayed flat, but rework dropped 12% - Your voice: "Boring meetings aren't the problem; bad feedback loops are." Section 3: "The Async Mindset Shift" (350 words, practical) - Async != slower; async = deliberate - Tools survey: Loom, Notion, GitHub comments (not Slack for decisions) - Contrarian take: Real-time communication is the enemy of deep work Conclusion: "3 experiments to run this quarter" - Measure pull-request feedback speed, not meeting attendance - Remove 1 recurring sync meeting; log productivity for 2 weeks - Audit: What % of your comms are "keeping people in the loop" vs "solving problems"? FULL DRAFT: 1,650 words, 8 citations, tone-matched to your voice Ready for: Light editing pass by you, then publish
About this skill
Your AI drafts cite studies you can't find. Their statistics trace back to a listicle quoting a press release. And the writing sounds nothing like you. This skill does the research properly, tracing each number to its original source and recording exactly what it measured. It then writes the piece in a voice built from checkable features of your past posts.
What it does
- Builds a voice fingerprint from 3-5 of your pieces: sentence rhythm, stance, openers and closers, how you attribute sources, words you never use, and signature moves (each capped so they don't turn into parody)
- Keeps a claim ledger before writing anything: each claim, its original source, the exact wording in the source, and a status (verified, secondhand, single-source, or unverified)
- Catches misquoted statistics, such as "tried once" turned into "uses daily," vendor numbers presented as research, viral figures with no traceable origin, and misattributed quotes
- Matches the strength of each verb to the evidence ("found" vs "suggests" vs "claims") and includes the strongest counter-evidence
- Works facts into the argument instead of listing them, then removes generic AI phrasing that doesn't match how you write
- Fact-checks and adds sources to a draft you paste in, or rewrites it to sound like you
What you get
A research brief, an outline, or a full draft, sized to what you asked for. Each comes with the claim ledger, a short voice fingerprint, and flags marking unverified claims and the places that need your own story or opinion.
Best for
Bloggers, newsletter writers, founders, and content marketers who publish evidence-backed long-form under their own name and can't afford a made-up citation.
SKILL.md preview
---
name: researched-article-writer-traced-citations-your-voice
description: Researches a topic and traces every claim to its original source in a claim ledger, then drafts a research brief, outline, or full article in the writer's own voice, using a fingerprint measured from their past work. Use when the user says "write a post on X in my voice with real sources", "research this and draft an article", "here are my last 3 posts, write the next one", "this draft sounds like AI, make it sound like me", "fact-check and source this draft", "I need stats for this piece but don't make anything up", "my AI drafts keep citing studies that don't exist".
---
… (sign up to view the full skill)More content & writing skills
View all Content & Writing skills →Newsletter Idea Bank and Next-Issue Writer
Ranked newsletter topic pitches tied to your readers' questions and your own material, then the winning issue drafted in your voice.
Client Case Study Writer with Numbers Check
Turns a client win into a credible case study after checking the math, attribution, quotes, and typical-results risks.
Publish-Ready Blog Post and Article Writer
Turns a topic, outline, or rough draft into a publish-ready blog post with a distinct angle, a fact-check list, and SEO metadata.
Quote Card Batch from Transcripts and Articles
Pulls exact, context-safe quotes from your episodes or articles and turns them into scored quote cards with captions, alt text, and a CSV.