メニュー
Design a source evaluation protocol using lateral reading and credibility checks for digital information. Use when students need to evaluate websites, online sources, or social media claims.
デフォルトでは、最初にソースを確認する Prompt が選択されています。直接コマンドに切り替えるか、ローカルコピーをダウンロードすることもできます。
インストールを決める前に、SKILL.md と SkillsMP に表示されている付属ファイルをお読みください。
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
直接コマンドでは確認用 Prompt が省略されます。実行前にソースを確認してください。
npx skills add https://github.com/GarethManning/education-agent-skills --skill source-credibility-evaluation-protocolコマンドは1行のまま表示されます。コピー前に横へスクロールして全体を確認してください。
ローカルで確認しますか？SkillsMP が現在取得できるファイルをダウンロードできます。
SOC 職業分類に基づく
SKILL.md を表示中
| name | source-credibility-evaluation-protocol |
| description | Design a source evaluation protocol using lateral reading and credibility checks for digital information. Use when students need to evaluate websites, online sources, or social media claims. |
| disable-model-invocation | false |
| user-invocable | true |
| effort | medium |
| skill_id | literacy-critical-thinking/source-credibility-evaluation-protocol |
| skill_name | Source Credibility Evaluation Protocol |
| domain | literacy-critical-thinking |
| version | 1.0 |
| evidence_strength | strong |
| evidence_sources | ["Wineburg & McGrew (2017) — Lateral reading: reading less and learning more when evaluating digital information","Wineburg & McGrew (2019) — Lateral reading and the nature of expertise","Breakstone et al. (2021) — Students' civic online reasoning: a national portrait","Caulfield (2019) — SIFT: the four moves (Stop, Investigate, Find better coverage, Trace claims)","Hobbs (2010) — Digital and media literacy: a plan of action"] |
| input_schema | {"required":[{"field":"source_type","type":"string","description":"The type of source students are evaluating — e.g. website, news article, social media post, Wikipedia article, YouTube video, infographic"},{"field":"evaluation_context","type":"string","description":"Why students are evaluating this source — the task or assignment context"},{"field":"student_level","type":"string","description":"Age/year group"}],"optional":[{"field":"specific_source","type":"string","description":"A description of the actual source being evaluated — topic, publisher, URL type"},{"field":"student_profiles","type":"array","description":"From context engine: digital literacy levels, prior experience with source evaluation"},{"field":"subject_area","type":"string","description":"The discipline context — affects what counts as credible evidence"},{"field":"common_mistakes","type":"array","description":"Source evaluation errors the teacher has observed in this class"}]} |
| output_schema | {"type":"object","fields":[{"field":"evaluation_protocol","type":"object","description":"Step-by-step protocol for evaluating this source type, using lateral reading"},{"field":"teacher_modelling","type":"string","description":"Script showing the teacher modelling the protocol with a specific example"},{"field":"red_flags_green_flags","type":"object","description":"Source-type-specific indicators of reliability and unreliability"},{"field":"student_checklist","type":"string","description":"A concise checklist students can use independently"}]} |
| chains_well_with | ["critical-thinking-task-designer","media-literacy-deconstruction-protocol","reading-comprehension-strategy-selector","socratic-questioning-sequence-generator","ai-hallucination-fact-check-protocol"] |
| teacher_time | 3 minutes |
| tags | ["source-evaluation","lateral-reading","SIFT","digital-literacy","critical-thinking"] |
Generates a structured lateral reading protocol for evaluating a specific type of source — teaching students to check what OTHER sources say about a source rather than analysing the source itself in isolation. This approach directly follows the research finding that professional fact-checkers outperform both students and university professors at source evaluation because they read laterally (opening new tabs to check who's behind a source) rather than vertically (reading the source itself more carefully for "clues" about reliability). The output includes a step-by-step protocol, a teacher modelling script, source-type-specific red and green flags, and a student checklist. AI is specifically valuable here because effective source evaluation requires both general principles (lateral reading, claim tracing) and source-type-specific knowledge (what makes a news article credible is different from what makes a scientific study credible) — a combination that is difficult to teach without domain expertise.
Wineburg & McGrew (2017, 2019) conducted landmark studies showing that professional fact-checkers evaluate sources fundamentally differently from students and even university professors. Fact-checkers use "lateral reading" — instead of staying on a source and looking for clues about its reliability, they immediately open new tabs to check what other sources say about the source's author, publisher, and claims. Students and professors, by contrast, use "vertical reading" — they stay on the page and look for surface credibility markers (professional design, .org domain, author credentials listed on the page) that are easily manipulated. Lateral readers took an average of 93 seconds to reach a correct evaluation; vertical readers took over 5 minutes and were more often wrong. Breakstone et al. (2021) found that the vast majority of US high school students cannot reliably evaluate online sources — they are easily deceived by professional-looking design and on-page credentials. Caulfield (2019) operationalised the research into the SIFT method: Stop (pause before engaging), Investigate the source (who's behind it?), Find better coverage (what do other sources say?), Trace claims (find the original source). This framework makes lateral reading teachable. Hobbs (2010) established that digital and media literacy requires structured instruction — students do not develop source evaluation skills naturally through internet use.
The teacher must provide:
Optional (injected by context engine if available):
You are an expert in digital literacy and source evaluation pedagogy, with deep knowledge of Wineburg & McGrew's (2017, 2019) research on lateral reading, Caulfield's (2019) SIFT method, and Breakstone et al.'s (2021) findings on students' civic online reasoning. You understand that most students (and many adults) evaluate sources using ineffective vertical reading strategies — checking the source itself for "clues" about reliability — and that effective evaluation requires lateral reading: checking what EXTERNAL sources say about the source in question.
Your task is to generate a source evaluation protocol for:
**Source type:** {{source_type}}
**Evaluation context:** {{evaluation_context}}
**Student level:** {{student_level}}
The following optional context may or may not be provided. Use whatever is available; ignore any fields marked "not provided."
**Specific source:** {{specific_source}} — if not provided, create a plausible example source of this type for the modelling script.
**Student profiles:** {{student_profiles}} — if not provided, assume students have limited formal training in source evaluation and tend to judge credibility by appearance (professional design = trustworthy).
**Subject area:** {{subject_area}} — if not provided, infer from the evaluation context.
**Common mistakes:** {{common_mistakes}} — if not provided, include the most common source evaluation errors for this age group: trusting professional-looking design, confusing .org with automatic credibility, accepting information because it confirms what they already believe.
Apply these evidence-based principles:
1. **Lateral reading over vertical reading (Wineburg & McGrew, 2017, 2019):**
- The protocol must teach students to LEAVE the source and check external references — not to scrutinise the source itself for credibility cues.
- Vertical reading cues students commonly rely on (and shouldn't): professional design, .org domain, author credentials listed on the page, presence of references, "About Us" page claims.
- Lateral reading moves: search the author's name, search the organisation's name + "credibility" or "funding," check if other reputable sources make the same claim, trace quoted statistics to their original source.
2. **SIFT framework (Caulfield, 2019):**
- **Stop:** Before engaging with the source, pause. Don't start reading, sharing, or citing until you've evaluated it.
- **Investigate the source:** Who is behind this? What do OTHER sources say about this author/organisation? (Not what the author/organisation says about themselves.)
- **Find better coverage:** Do other reliable sources report the same information? If only one source makes this claim, be cautious.
- **Trace claims:** When a source cites a statistic, quote, or study, trace it back to the original. Does the original say what the source claims it says?
3. **Source-type-specific guidance:**
- Different source types have different credibility indicators and different manipulation techniques.
- A protocol for evaluating a news article should address different issues than one for evaluating a social media infographic.
- Adapt the SIFT steps to the specific source type.
4. **Avoid false sophistication:**
- Students often learn a checklist approach (CRAAP test, etc.) that produces a false sense of sophistication — they check boxes without actually evaluating. The protocol should teach THINKING, not box-ticking.
- The most important question is always: "What do INDEPENDENT, KNOWLEDGEABLE sources say about this?" — not "Does this source LOOK reliable?"
5. **Model the process explicitly:**
- Include a think-aloud modelling script where the teacher demonstrates lateral reading with a specific example.
- Show the contrast between what a vertical reader would do (and get wrong) and what a lateral reader would do (and get right).
Return your output in this exact format:
## Source Evaluation Protocol: [Source Type]
**For:** [Student level]
**Context:** [Why students are evaluating this source]
**Method:** Lateral reading (SIFT framework)
### The Protocol
[Step-by-step protocol adapted for this source type — each step should explain WHAT to do, WHY it matters, and WHAT TO LOOK FOR]
### Teacher Modelling Script
[A think-aloud script showing the teacher evaluating a specific example of this source type, demonstrating lateral reading. Show the contrast with what a vertical reader might do wrong.]
### Red Flags and Green Flags
**Red flags (reasons to be cautious):**
[Source-type-specific indicators of unreliability]
**Green flags (reasons to be more confident):**
[Source-type-specific indicators of reliability]
**Important: green flags are not proof.** [Brief note that even reliable-looking sources can be wrong — green flags increase confidence but don't guarantee accuracy]
### Student Checklist
[A concise, printable checklist students can use independently when evaluating sources of this type]
**Self-check before returning output:** Verify that (a) the protocol prioritises lateral reading over vertical reading, (b) each step teaches thinking, not box-ticking, (c) the modelling script demonstrates the actual process of checking external sources, (d) red and green flags are specific to this source type, (e) the protocol is age-appropriate, and (f) the student checklist is concise enough to be practically usable.
Scenario: Source type: "Website" / Evaluation context: "Year 9 students are researching the health effects of energy drinks for a Science project. Several students have found websites claiming energy drinks are safe or even beneficial, and the teacher wants to teach them how to evaluate these sources before including them in their projects." / Student level: "Year 9"
For: Year 9 Science Context: Evaluating websites found during research on health effects of energy drinks Method: Lateral reading (SIFT framework)
Step 1: STOP — Don't start reading yet (30 seconds)
Before you read a single word of the article, pause. You've found this website through a search engine — but appearing in search results does NOT mean a source is reliable. Google ranks pages by popularity and relevance to your search terms, not by accuracy.
What to do: Note the URL. Note the name of the website/organisation. Do NOT start reading the article yet.
Why this matters: Once you start reading, your brain begins accepting the information. It's much harder to evaluate a source critically after you've already absorbed its claims. Evaluate FIRST, read SECOND.
Step 2: INVESTIGATE THE SOURCE — Who is behind this website? (2–3 minutes)
This is the critical step. Do NOT check the website's own "About Us" page — any organisation can write flattering things about itself. Instead, OPEN A NEW TAB and search externally.
What to do:
What to look for:
Step 3: FIND BETTER COVERAGE — Do other reliable sources say the same thing? (2–3 minutes)
If the website makes specific claims ("Energy drinks have no negative health effects on teenagers"), check whether reputable sources agree.
What to do:
What to look for:
Step 4: TRACE CLAIMS — Where do the statistics come from? (2–3 minutes)
If the website cites numbers, studies, or expert quotes, trace them back to the original source.
What to do:
What to look for:
"I'm going to show you how I'd evaluate a website I found while searching for information on energy drinks. I've found a website called 'EnergyDrinkFacts.org' — it looks professional, has a clean design, and has an article titled 'The Truth About Energy Drinks and Teen Health.' Let me show you what most people do — and then what I do differently.
What most people do (vertical reading — this is the WRONG approach): Most people would start reading the article. They'd see the .org domain and think 'this must be legitimate.' They'd see that the article cites some studies and think 'this is evidence-based.' They'd see the professional design and think 'this looks trustworthy.' They might check the About Us page and read 'We are an independent research organisation dedicated to providing factual information about energy beverages.' Sounds good, right? This is vertical reading — staying on the website and looking for clues. And it's exactly how misinformation works. ANYONE can register a .org domain. ANYONE can make a professional-looking website. ANYONE can write a nice About Us page.
What I do instead (lateral reading — the RIGHT approach): I'm going to LEAVE this website immediately and open a new tab.
[Opens new tab] I search: 'EnergyDrinkFacts.org' — and I find... interesting. The Wikipedia article says this organisation was founded and funded by the American Beverage Association — the trade body that represents energy drink manufacturers. So the 'independent research organisation' is actually funded by the companies that sell energy drinks. That's not automatically disqualifying, but it's ESSENTIAL context that I would never have found by reading the website itself.
Now I search: 'energy drinks health effects teenagers NHS' — and the NHS page says caffeine in energy drinks can cause anxiety, sleep disruption, and heart palpitations in teenagers, and advises against regular consumption. The WHO says the same. So the claim on EnergyDrinkFacts.org that energy drinks are safe for teens CONTRADICTS what the NHS and WHO say.
Now I trace a specific claim. The website says 'A 2018 study found no significant health risks from moderate energy drink consumption.' I search for this study... and I find it was funded by Red Bull and published in a journal with a low impact factor. The larger, independent studies published in mainstream medical journals found the opposite.
My conclusion: this website LOOKS professional and credible, but lateral reading reveals it's industry-funded and its claims contradict the independent evidence. I would NOT use this as a source in my project.
That whole process took me about 3 minutes. And I barely read the article itself — because the article isn't where the truth is. The truth is in what INDEPENDENT sources say."
Red flags (reasons to be cautious):
Green flags (reasons to be more confident):
Important: green flags are not proof. Even established institutions can publish errors, and reliable-looking sources can be wrong on specific claims. Green flags increase confidence; they don't guarantee accuracy. Always check specific claims against other sources.
SIFT Check — Use this when evaluating any website for your project
| Step | Action | ✓ |
|---|---|---|
| STOP | I have NOT started reading the article yet. I will evaluate first. | |
| INVESTIGATE | I searched the organisation's name in a new tab. I know who is behind this website. | |
| INVESTIGATE | I searched "[organisation] funding" or "backed by." I know who funds them. | |
| INVESTIGATE | The organisation does / does not have a financial interest in this topic. | |
| FIND | I searched the main claim in a new tab. Other reputable sources agree / disagree / I can't find any other source making this claim. | |
| TRACE | I found a specific statistic or claim and traced it to the original source. The original source does / does not support what the website says. | |
| MY VERDICT | Based on lateral reading, I rate this source as: Reliable / Use with caution / Do not use | |
| MY REASONING | I reached this verdict because: _________________________________ |
Lateral reading requires internet access and time. The SIFT process takes 3–5 minutes per source, which is manageable for a research project but impractical for every piece of information students encounter online. The skill teaches a thorough evaluation process best suited for academic research tasks — students also need quicker heuristics for everyday digital literacy (e.g., "check the source before sharing"), which this protocol doesn't cover.
The protocol teaches evaluation of individual sources but not synthesis across sources. A student who can evaluate one website may still struggle to synthesise information from multiple sources, weigh conflicting evidence, or recognise that even credible sources can disagree. Source evaluation is a necessary but not sufficient skill for research competence.
Some source types are harder to evaluate laterally than others. Websites and news articles are relatively straightforward to check — organisations and authors can be searched. Social media posts, anonymous forum contributions, and viral content shared without attribution are much harder to trace. The protocol is most effective for sources with identifiable authors and publishers.