Source-Grounded Research Brief (Claims, Evidence, Unknowns)
This prompt organizes research into claims, source evidence, dates, and unresolved questions. It prioritizes primary documentation and keeps a clear boundary between what a source states and what the writer infers.
GPT / Claude / Gemini2 variables
Prompt
Create a source-grounded research ledger for {TOPIC}.
Sources supplied:
{SOURCES}
For each material claim, provide:
- Claim in plain language
- Supporting source and exact URL
- Short evidence excerpt or section reference, if supplied
- Source type: primary, secondary, or unverified
- Publication/update date and date checked, if known
- What the source does not establish
- Confidence: supported, partially supported, conflicting, or unverified
- Next verification step
Rules:
- Never invent URLs, quotations, dates, prices, statistics, or source contents.
- Prefer official documentation, release notes, standards, and first-party announcements.
- Keep facts separate from interpretation.
- Flag conflicting sources and stale information.
- If the supplied evidence is insufficient, write "not established by the available sources."
Topic: {TOPIC}Quick brief
Purpose
Build a research trail that makes factual writing easier to verify.
Expected output
A claim ledger with claim text, source, evidence excerpt or location, source type, date checked, confidence, caveats, and next verification step.
Customize before copying
Replace these placeholders with your own context before you run the prompt.
{TOPIC}{SOURCES}
Works well with
GPT
Claude
Gemini
Variations
Build a price-and-feature verification table for a software comparison.
Audit a release article for claims that lack primary-source evidence.
Convert a research ledger into an editor-friendly fact-check checklist.
What this prompt helps you do
This prompt organizes research into claims, source evidence, dates, and unresolved questions. It prioritizes primary documentation and keeps a clear boundary between what a source states and what the writer infers.
When to use it
Use for product documentation, AI model updates, technical explainers, pricing changes, policy summaries, and any article where stale or misattributed information could mislead readers.
Decision context
Research depth should match the risk and volatility of the topic. A stable definition may need a small number of authoritative references; a current product comparison needs dated source checks for each changing feature or price.
How it works
The model extracts individual claims, links each to a supplied source, records the source date where available, and labels gaps. It does not treat a search snippet, affiliate page, or unsourced summary as equivalent to an official announcement.
Best practices
Provide direct source URLs and relevant excerpts. Capture access dates for volatile details. Keep exact quotations short and attribute them. Verify important claims in the source itself rather than relying on a generated summary.
Common mistakes
Never ask the model to fill missing citations, infer a launch date from rumors, or turn an unverified claim into confident prose. A source existing does not mean it supports every claim nearby.
What you should expect back
A claim ledger with claim text, source, evidence excerpt or location, source type, date checked, confidence, caveats, and next verification step.
Limitations
The prompt cannot open inaccessible links or independently establish that a source is authentic. Human review remains necessary for material claims and changing facts.
Model notes
Supply source material directly if browsing is unavailable. For current information, use the official vendor documentation or announcement wherever possible.
Real-world applications
Useful for model-release notes, SaaS feature pages, technical tutorials, purchasing guides, and fact-checking existing content.
How to tell if it worked
Every material claim has a traceable basis, time-sensitive facts are dated, unknowns remain visible, and a reviewer can reproduce the verification quickly.
Where to go next
Use before the tutorial writer or content quality audit.
Appears in collections
Related prompts
SEO Content Brief (Search Intent to Editorial Plan)
Turn a topic into a useful, evidence-led brief before drafting.
User Research Synthesis (Patterns to Insights)
Transform raw user research data into actionable insights and recommendations.
Survey Design Framework (Insights Without Bias)
Design surveys that collect actionable insights while avoiding common bias and quality issues.
Audience Persona Builder (Research → Action)
Create detailed audience personas grounded in real research, not assumptions, so marketing and product decisions are data-driven.
Content Quality Audit (Originality, Accuracy, Reader Value)
Review a draft for usefulness and credibility before publishing.