Fill in all {{PLACEHOLDER}} variables below before pasting into any AI tool. Each placeholder is described in the Variable Reference section. Remove any categories or patterns that don't apply to your investigation. The template works across any platform (Twitter/X, Facebook, Instagram, Telegram, YouTube, TikTok, etc.) and any language or country context.
| Placeholder | What to enter | Example |
|---|---|---|
{{ACCOUNT_HANDLE}} |
Account username/handle to investigate | @someaccount |
{{PLATFORM}} |
Primary platform being analyzed | Twitter/X |
{{PLATFORM_1}} {{PLATFORM_2}} {{PLATFORM_3}} |
Other platforms where the account is present | Facebook, Instagram, Telegram |
{{COUNTRY}} |
Country or region context | Brazil |
{{TIMEFRAME_MONTHS}} |
How many months back to analyze | 12 |
{{EVENT_PERIOD_1}} {{EVENT_PERIOD_2}} |
Specific date ranges for timing analysis (Category 3) | March 2024 elections, June 2024 protests |
{{LEGAL_EVENT}} |
Legal action, lawsuit, or regulatory event affecting the outlet | 2023 defamation lawsuit filed by Party X |
{{EVENT_NAME}} |
Name of the critical event for deep-dive (Pattern A) | Mayoral election scandal |
{{EVENT_DATE_RANGE}} |
Date range for the critical event | April 1-15, 2024 |
{{NUMBER}} |
How many events to compare across outlets (Pattern B) | 5 |
{{OUTLET_1}} {{OUTLET_2}} {{OUTLET_3}} |
Names of comparison media outlets | BBC, Reuters, AFP |
{{OUTLET_1_DESCRIPTION}} {{OUTLET_2_DESCRIPTION}} {{OUTLET_3_DESCRIPTION}} |
Brief descriptor of each outlet's editorial stance | center-left national daily, state-owned broadcaster, independent investigative |
{{CORRECTION_KEYWORDS}} |
Comma-separated keywords in the account's language for finding corrections/retractions | correction, retraction, clarification, updated, we regret |
{{CROSS_PLATFORM_SEARCH_TERMS}} |
Search terms for finding discussions about content differences across platforms | "different on [platform]", "deleted from [platform]", "not on [platform]" |
{{CRISIS_EVENT}} |
Name of the crisis period for coverage gap analysis (Pattern E) | Mass protests |
{{CRISIS_DATE_RANGE}} |
Date range for the crisis period | October 1-30, 2024 |
CORE QUESTION: Act as an OSINT analyst specializing in media credibility and influence operations. Analyze the {{ACCOUNT_HANDLE}} account on {{PLATFORM}} for editorial consistency and potential manipulation. Your investigation must cover all 8 categories below, all 5 special investigation patterns, and produce a final verdict with evidence.
INITIAL OUTPUT: Begin with a one-paragraph verdict answering: Is this account a neutral news aggregator, a biased outlet, or an active manipulation vector? Include confidence level (X/100).
Analyze whether the account frames events consistently or applies different framing depending on the political actor involved.
Sub-questions:
- Do headlines match actual content? Find examples where the headline/summary diverges from the article or linked source.
- Is the same event type framed differently depending on which political actor is involved? Identify at least two events of similar nature involving different actors and compare the framing.
- Compare how the account covers government figures vs opposition figures for the same type of event (e.g., corruption allegations, policy announcements, public appearances).
Output format:
CATEGORY | FINDINGS | VERDICT | CONFIDENCE
Tag each finding as [Fact] or [Inference] or [Unverified]. Confidence is a number/100. Cite specific posts with dates and engagement numbers. Note edge cases at the end.
Analyze what the account chooses to cover and what it chooses to ignore. Selection bias reveals editorial agenda more reliably than how something is covered.
Sub-questions:
- What topics are consistently amplified? Map the top 5-10 most frequently covered topics over the last
{{TIMEFRAME_MONTHS}}months. Are they disproportionately focused on a specific political pole, region, or narrative? - What topics are consistently ignored or underreported? Identify major events that received widespread media coverage but were absent or minimally covered by this account. Search specifically for events that would contradict the account's apparent narrative.
- Is there a pattern in which sources they cite vs avoid? Do they consistently reference a particular set of outlets while ignoring others? Map the source citation network (who they cite → what those sources represent).
Output format:
CATEGORY | FINDINGS | VERDICT | CONFIDENCE
Tag each finding as [Fact] or [Inference] or [Unverified]. Confidence is a number/100. Cite specific posts with dates and engagement numbers. Note edge cases at the end.
Examine whether the account's posting behavior correlates with political events in ways that suggest coordination or agenda-driven amplification.
Sub-questions:
- Do breaking news posts cluster around politically sensitive moments? Map posting frequency against a timeline of major political events. Look for volume spikes that coincide with events unfavorable to one side or favorable to another.
- Is there unusual posting activity during elections, protests, or crises? Compare hourly/daily posting rates during these periods vs baseline periods.
- Investigate posting behavior during:
{{EVENT_PERIOD_1}},{{EVENT_PERIOD_2}}, and any major{{COUNTRY}}political events in the last{{TIMEFRAME_MONTHS}}months.
Output format:
CATEGORY | FINDINGS | VERDICT | CONFIDENCE
Tag each finding as [Fact] or [Inference] or [Unverified]. Confidence is a number/100. Cite specific posts with dates and engagement numbers. Note edge cases at the end.
Analyze whether engagement metrics appear organic or suggest coordinated amplification — either through bot networks, engagement pods, or inauthentic coordinated behavior.
Sub-questions:
- Is engagement (likes, retweets/shares, comments) organic relative to follower count? Calculate engagement rate (average engagement per post / follower count) and compare against platform benchmarks for accounts of similar size.
- Are there posts with abnormally high or low engagement vs the account's average? Identify outliers — posts that massively overperform or underperform relative to the account's typical engagement baseline. Investigate what made those posts different.
- Do specific post types get amplified by coordinated accounts? Examine the accounts that most frequently retweet/share or comment on this account's posts. Check for: accounts created around the same time, accounts with similar naming patterns, accounts that only interact with this account, accounts that all engage with the same posts within a narrow time window.
Output format:
CATEGORY | FINDINGS | VERDICT | CONFIDENCE
Tag each finding as [Fact] or [Inference] or [Unverified]. Confidence is a number/100. Cite specific posts with dates and engagement numbers. Note edge cases at the end.
Evaluate the account's journalistic standards regarding source transparency, attribution practices, and willingness to issue corrections.
Sub-questions:
- Do they credit original sources? Sample 20-30 posts and check whether the original source of information is clearly named and linked. What percentage of posts attribute properly?
- Are sources verifiable or vague? Count and categorize source language: specific named sources vs "sources say" vs "according to reports" vs "it is claimed that" vs no source at all. Vague attribution is a hallmark of disinformation laundering.
- Do they ever correct or retract stories? Search the account's history for any posts acknowledging errors. If none exist over a long period despite known controversial posts, this is itself a finding.
Output format:
CATEGORY | FINDINGS | VERDICT | CONFIDENCE
Tag each finding as [Fact] or [Inference] or [Unverified]. Confidence is a number/100. Cite specific posts with dates and engagement numbers. Note edge cases at the end.
Check whether the account tailors its messaging to different audiences across platforms, which can reveal which content is intended for which demographic.
Sub-questions:
- Is messaging on
{{PLATFORM_1}},{{PLATFORM_2}}and{{PLATFORM_3}}consistent? Compare posts on the same topic across platforms. Look for differences in tone, framing, completeness of information, or target audience. - Are there posts on one platform that contradict or disappear from another? Search for documented cases where content existed on one platform but not on others, or where the same event was described differently across platforms.
Output format:
CATEGORY | FINDINGS | VERDICT | CONFIDENCE
Tag each finding as [Fact] or [Inference] or [Unverified]. Confidence is a number/100. Cite specific posts with dates and engagement numbers. Note edge cases at the end.
Examine how external pressure — legal, regulatory, or investigative — has affected the account's behavior.
Sub-questions:
- How did the account respond to the
{{LEGAL_EVENT}}? Analyze posting patterns, editorial tone, and content selection before, during, and after the legal pressure. Did the account go silent, become more aggressive, shift topics, or change tone? - Did editorial tone or content selection change after legal pressure? Compare the 3 months before vs 3 months after the legal event. Look for measurable shifts in topic distribution, sentiment, or sourcing practices.
- Search the web for any court documents, regulatory actions, or journalist investigations about the outlet. Gather all publicly available legal or regulatory records involving the account, its operators, or its associated organization.
Output format:
CATEGORY | FINDINGS | VERDICT | CONFIDENCE
Tag each finding as [Fact] or [Inference] or [Unverified]. Confidence is a number/100. Cite specific posts with dates and engagement numbers. Note edge cases at the end.
Map the account's position within the broader information ecosystem — who amplifies it and who it amplifies in return.
Sub-questions:
- Which political accounts or media outlets most frequently amplify this account? Identify the top 10-20 accounts that retweet/share, quote, or otherwise redistribute this account's content. Categorize them by type: political figures, partisan media, activists, anonymous accounts, bot-like accounts.
- Is there overlap between the account's amplification network and known influence operation accounts? Cross-reference the amplification network against any publicly documented influence operations, coordinated inauthentic behavior networks, or state-linked media entities.
- Does the account engage with or avoid specific political figures? Map interaction patterns (replies, mentions, retweets/shares) with key political figures. Does it engage positively with some and negatively with others? Does it systematically avoid mentioning certain figures?
Output format:
CATEGORY | FINDINGS | VERDICT | CONFIDENCE
Tag each finding as [Fact] or [Inference] or [Unverified]. Confidence is a number/100. Cite specific posts with dates and engagement numbers. Note edge cases at the end.
Top 3 red flags (if any), and top 3 credibility indicators. Cite specific posts with dates where possible. Re-state the overall verdict from the initial output, now supported by the evidence from all 8 categories.
The following deep-dive patterns must be applied in addition to the 8 categories above.
For the specified critical event period ({{EVENT_NAME}}, {{EVENT_DATE_RANGE}}):
- How many posts were made about the event vs normal daily volume? Calculate the percentage of total posts during this period that were about the event. Compare to the account's average daily post volume.
- What framing was used? Categorize each post about the event as: legal process framing, political attack framing, neutral reporting, sympathetic framing, or other. Show the distribution.
- Which sources were cited? List every source cited during event coverage and categorize by type: official government sources, opposition sources, other media outlets, anonymous sources, no source.
- Compare: did the account cover all sides of the event? List which stakeholders' perspectives were represented. If some sides were absent, note this as a finding.
- If yes, how were different sides framed? Compare the language, adjectives, and emotional tone used for each side. Is there symmetry or asymmetry?
- Were any posts deleted or edited in this period? Check for posts that were removed or significantly altered. If found, document: original content, timing of deletion/edit, whether a correction was posted, and what the edited version changed.
- Show specific post examples with dates and engagement numbers for each finding above.
Pick {{NUMBER}} major political events from the last {{TIMEFRAME_MONTHS}} months that the account covered.
For each event, compare how the account framed it vs:
{{OUTLET_1}}({{OUTLET_1_DESCRIPTION}}){{OUTLET_2}}({{OUTLET_2_DESCRIPTION}}){{OUTLET_3}}({{OUTLET_3_DESCRIPTION}})
For each comparison, produce a table:
OUTLET | HEADLINE/FRAMING | SOURCE CITED | EMOTIONAL TONE | WORD CHOICE DIFFERENCES
After completing all event comparisons, conclude: Does the account consistently align with any single outlet's framing? If so, which one and with what strength? If not, does it align with a particular ideological position that spans outlets?
Search the account for posts containing correction-related keywords: {{CORRECTION_KEYWORDS}}.
For each correction/retraction found:
- What was the original claim? Document the exact text, date, and engagement numbers of the original post.
- How quickly was the correction made? Calculate the time delta between original post and correction. Categorize: within 1 hour, within 24 hours, within 1 week, longer.
- Did the correction get similar engagement to the original post? Compare likes, retweets/shares, and comments on the correction vs the original. A correction that gets 5% of the original's reach is functionally invisible.
- Was the original misleading post deleted or kept? Check whether the original post remains accessible. If it was deleted without a visible correction, that is a separate finding. If it was kept but the correction was a reply or separate post, that is also a finding.
This analysis reveals whether corrections are genuine editorial practices or damage-control optics designed to create a defense ("we corrected it") while ensuring the misleading information continues to spread.
If no corrections are found despite a long posting history, note this as a finding: the account either never makes errors (statistically improbable for high-volume accounts) or never acknowledges them.
Search for posts and discussions about the account's content differences across platforms.
Also search: {{CROSS_PLATFORM_SEARCH_TERMS}}
Then investigate:
- Are there any documented cases where content appeared on one platform but NOT on another? Look for: posts that exist on one platform but are absent on another during the same time period; content that was posted and then removed from one platform but remains on another; content that contradicts between platforms.
- What does the bio/profile description on each platform say vs others? Compare the account's self-description, listed affiliations, links, and pinned/featured content across all platforms. Differences may indicate audience segmentation or attempts to appear neutral on some platforms while being openly partisan on others.
- Are pinned/highlighted messages on one platform politically different from another? Compare the content of pinned posts, featured stories, or highlighted content across platforms for differences in political tone, topic selection, or framing.
For the {{CRISIS_EVENT}} period ({{CRISIS_DATE_RANGE}}):
- Which accounts on
{{PLATFORM}}DID cover the event directly with high volume? Identify the top 10-20 accounts that posted most frequently about the crisis during the specified period. Categorize them by type: news media, activists, eyewitnesses, political figures, OSINT accounts. - Did the target account retweet or quote any of those accounts during this period? Check whether
{{ACCOUNT_HANDLE}}interacted with any of the high-volume crisis coverage accounts. If it did not, this is a finding: the account was active but chose not to engage with or amplify coverage of the crisis. - Search for: accounts that mentioned the target account AND the crisis topic in the same post during this period. What were they saying? Look for: accusations of silence, calls for the account to cover the event, criticism of the account's framing if it did cover it, or praise if the account covered it in a particular way.
- Did the target account's follower count change significantly during or after this period? Check follower growth/loss patterns. A significant change during a crisis — especially a loss — may indicate audience reaction to coverage choices or lack of coverage.
- Were there any posts criticizing the target account for NOT covering the event? Search for posts that name the account alongside complaints about silence, censorship, selective coverage, or agenda-driven omission.
Apply these formatting rules throughout the entire analysis:
-
Each of the 8 main categories must produce a summary table:
CATEGORY | FINDINGS | VERDICT | CONFIDENCEWhere FINDINGS is a bulleted list, VERDICT is a 1-2 sentence assessment, and CONFIDENCE is a number out of 100.
-
Each finding must be tagged with exactly one of:
[Fact],[Inference], or[Unverified].[Fact]: Directly observable from the data (e.g., "Posted 47 times on March 15")[Inference]: Reasonable conclusion drawn from multiple facts (e.g., "Posting pattern suggests coordination with Event X")[Unverified]: Plausible but not confirmable with available data (e.g., "May be linked to Network Y based on overlapping followers")
-
Post examples must be cited with: post date, a brief description of content, and engagement numbers (likes, retweets/shares, comments).
-
Edge cases must be noted at the end of each section. An edge case is any finding that complicates the overall assessment or doesn't fit neatly into the pattern — for example, a post that breaks the account's usual framing pattern, or an engagement outlier with an ambiguous explanation.
-
All special investigation patterns (A through E) must be completed with their full sub-analysis structure as specified.
-
The final verdict must include top 3 red flags and top 3 credibility indicators, each cited with specific post dates.