An AI skill is a structured system prompt or set of custom instructions that gives AI assistants like ChatGPT, Claude, and Gemini specialized capabilities. It includes a clear role definition, specific knowledge areas, step-by-step processes, and quality criteria that help AI perform specific professional tasks.

How does Skill This work?

Simply describe your professional expertise in plain text. Our free AI skill generator analyzes your input and transforms it into a structured system prompt. The skill is then graded on clarity, specificity, and usefulness. You can share your skill via a unique URL or copy it to use as custom instructions with any AI assistant like ChatGPT, Claude, or Gemini.

Is Skill This free to use?

Yes, Skill This is completely free. No account or payment is required. You can create unlimited skills and share them with anyone.

How is my skill graded?

Skills are graded on a scale from A+ to F (0-100 points). The grade reflects how clear, specific, and actionable your skill is. Higher grades indicate skills that AI assistants can use more effectively.

How do I use my skill with an AI assistant?

Copy your generated skill and paste it into the system prompt or custom instructions of any AI assistant. For ChatGPT, use Personalization > Custom Instructions. For Claude, add it to a Project's instructions. For Cursor or other IDEs, save it as a .cursorrules or CLAUDE.md file. The AI will then adopt that skill and perform tasks according to your expertise.

AI Skill Report Card

Creating AI Assisted Investigative Journalism

B+76·Nov 8, 2026·Source: Web

Creating AI-Assisted Investigative Journalism

Quick Start11 / 15

Given a topic and deadline, run the pipeline: research → structure → draft → fact-check → edit → multi-format adaptation → publish → measure. Example invocation:

Topic: "AI's impact on labor market"
Deadline: 3 weeks
Formats: longread + infographics + podcast
→ Output: 10-article series, 15 infographics, 1 podcast episode, social cuts

Recommendation▾

Quick Start example shows only input/output shape, not actual generated content (e.g., a sample angle, headline, or draft excerpt) — make it more concretely illustrative of real output quality.

Workflow14 / 15

Progress:

Step 1: Formulate topic and angle (what's new, why now, who cares)
Step 2: Research via AI (Perplexity/Elicit) — collect and triage 100+ sources
Step 3: Generate structure and draft (Claude/GPT-4o)
Step 4: Fact-check and verify every claim and number against primary sources
Step 5: Edit for style, accuracy, and ethics (Grammarly/LanguageTool + human pass)
Step 6: Adapt for platforms (longread, social snippets, podcast script)
Step 7: Publish and track engagement metrics

Step 1: Topic & Angle

Define the core tension or surprise (what contradicts common assumptions)
Write a one-sentence thesis before researching — this prevents aimless data collection
Identify 2-3 sub-angles for a series (each becomes one article)

Step 2: Research & Analysis

Use Perplexity for broad source discovery (aim for 200-400 sources on a major series)
Use Elicit for academic paper synthesis and extracting structured findings
Tag sources by reliability tier (primary data, expert interview, secondary reporting, speculation)
Flag contradictions between sources explicitly — these often become story angles themselves

Step 3: Structure & Draft Generation

Generate a series-level outline first (how articles connect, what's deferred to which piece)
Draft article-by-article with Claude/GPT-4o, feeding it the verified research tier, not raw search results
Keep drafts in distinct voice-set prompts per section (data-heavy vs. narrative vs. human-interest)

Step 4: Fact-Check & Verification

Every number, quote, and statistic gets a source citation before moving to editing
Cross-check AI-generated summaries against original source text — AI paraphrase can subtly shift meaning
Use a second AI pass specifically prompted to "find unsupported claims in this draft"

Step 5: Editorial Pass

Run Grammarly/LanguageTool for mechanical issues first (cheap pass before human time)
Human/AI review for: tone consistency across the series, ethical framing (avoiding sensationalism, protecting sources), narrative balance
Check that AI didn't flatten nuance or inject false balance on settled facts

Step 6: Multi-Platform Adaptation

Longread: full narrative with embedded data visualizations (Datawrapper/Flourish)
Podcast: convert key sections into conversational script, record/edit in Descript
Social: extract 3-5 standalone insights per article as short-form posts with visual hooks

Step 7: Publish & Analyze

Stagger publication across the series to sustain attention (not all at once)
Track per-format engagement separately (longread completion rate, infographic reposts, podcast listen-through)
Feed engagement data back into angle selection for the next series

Recommendation▾

The two examples are solid but still somewhat summarized/abstract (stats-heavy) rather than showing actual text snippets like a generated headline list, a sample angle framing, or a fact-check flag — add one example with real sample text output.

Where AI Creativity Matters Most (and How to Direct It)

This is the core skill: AI isn't just an execution tool — its creative judgment, applied correctly, determines whether output is "generic AI content" or genuinely insightful journalism.

Angle generation — ask AI to propose 10 contrarian or underexplored angles on a topic, not just the obvious one. Select for angles that create productive tension, not just novelty.
Analogy and framing invention — AI should generate multiple explanatory analogies for complex concepts (e.g., explaining labor market displacement) and you pick the one that resonates with the target audience, not the first one generated.
Structural experimentation — ask AI to propose 2-3 different narrative structures for the same material (chronological, thematic, character-driven) before committing.
Synthesizing contradictions into narrative tension — when sources disagree, AI's creative task is to frame the disagreement as the story's engine, not smooth it over.
Visual/format translation — AI should creatively propose which data points are visual-native (charts) vs. narrative-native (anecdote) vs. audio-native (interview-style), not just reformat the same content three times.
Headline and hook variation — generate 10-15 headline variants per piece, explicitly optimizing for different reader psychology (curiosity gap, stakes, surprise), then A/B or editorially select.
Voice differentiation across series — if writing 10 articles, deliberately vary rhetorical approach per piece (one data-led, one profile-led, one historical-context-led) so the series doesn't read as repetitive AI output.
Audience empathy simulation — prompt AI to anticipate reader objections/questions and preemptively address them in text — this is a creative inference task, not retrieval.
Cross-domain borrowing — explicitly prompt AI to borrow structural techniques from other genres (true crime pacing, scientific paper logic, oral history) to avoid formulaic journalism templates.
Ethical creativity — the hardest form: creatively finding ways to represent vulnerable subjects/sources with dignity while maintaining narrative power, not defaulting to either sanitized blandness or exploitative drama.

Examples14 / 20

Example 1: Input: 10-article series on AI's impact on labor market, 3-week deadline, longread + infographic + podcast Output: Perplexity analyzed 340 sources → Claude built series structure and drafts → fact-checking pass → editorial pass → Datawrapper produced 15 infographics → Descript produced 45-min podcast → social adaptation. Result: 10 articles in 18 days, 240,000 views, 1,200 infographic reposts, 45,000 podcast listens, series won a journalism award.

Example 2: Input: Single investigative longread on a local housing policy, 5-day deadline, no podcast Output: Perplexity + Elicit gather ~80 sources and 3 academic studies → AI proposes 5 angles, editor selects "policy helps homeowners but accelerates renter displacement" tension → draft generated with that framing → fact-check against city records → 2 infographics via Flourish → publish with 5 social pull-quotes highlighting the contradiction.

Recommendation▾

The 'Where AI Creativity Matters' section is valuable but lengthy (10 numbered points) and could be tightened or merged with Best Practices to reduce redundancy and improve conciseness.

Best Practices

Always separate "AI-researched" from "AI-verified" — never publish a claim that only passed through the research stage
Generate multiple creative options (angles, headlines, structures) before selecting — single-shot AI output is rarely the best option
Match format to content nature, don't force every data point into every format
Keep a visible citation trail from draft back to primary source for every factual claim
Deliberately vary AI prompting style across a series to avoid homogeneous "AI voice" fatigue in readers

Common Pitfalls

Treating AI's first draft as final — the creative value is in iteration and selection, not first output
Skipping fact-check on AI-synthesized summaries, assuming the underlying sources were accurate
Using AI only for mechanical tasks (editing, formatting) while missing its value in angle/structure ideation
Publishing a series where every article has identical structure and pacing — a sign AI wasn't prompted for structural variation
Letting AI smooth over source contradictions instead of using them as narrative tension
Forgetting ethical review — AI has no inherent judgment about source protection or sensitive representation