← Back to all workflows
Analysis

Do you want to analyze an important email?

This workflow performs a sentence-by-sentence dissection of an email, including rephrasing, grammar checks, fact-checking, credibility assessment, rhetorical analysis, and an effectiveness score.

Under the Hood

This is the actual text of the workflow

Email Analysis Workflow

When the user provides an email (or email chain) for analysis, perform the following steps. If the email is part of a chain, analyze only the most recent email unless told otherwise, but use the full chain as context.

I. Sentence-by-Sentence Breakdown

For each sentence in the email, provide the breakdown below. Every sentence must be analyzed individually — never collapse multiple sentences into a single block entry, even if they form a block quote, a proposed script, or a thematically unified passage. If the email contains a multi-sentence quoted passage, break it into its component sentences and analyze each one.

1. Original

Write out the sentence exactly as it appears.

2. In My Own Words

Rephrase the sentence in clear, plain language to demonstrate comprehension. Speak in the sender's voice (first person), as if you are the sender restating your own point more clearly — do NOT commentate about the sender in third person (e.g., "The sender is trying to ..." is wrong; "I'm saying that ..." is right).

3. Claim Decomposition

Decompose the sentence into every discrete claim it contains, listing each separately (e.g., "Fact: X", "Opinion: Y"). CRITICAL: Decompositions, fact-checks, and credibility assessments must evaluate the exact text and scope of the original claim, not your own rephrased "In My Own Words" version. Never render a single combined verdict across mixed claim types.

4. Grammar / Spelling / Punctuation / Word Usage

  • Flag genuine errors only (typos, missing punctuation, incorrect capitalization, garbled phrasing).
  • Bold only actual errors in the final output. If a sentence is clean, say so briefly and move on.
  • Do NOT flag stylistic choices as errors. For example, a long sentence spliced together with commas may be technically improper punctuation but can still be perfectly easy to follow — judge by actual comprehension, not by grammar textbook rules.
  • Check whether any words are used incorrectly (i.e., the writer didn't understand what the word means). Note: creative or unconventional word usage that is clearly intentional (e.g., using "graffitis" as a verb) is not an error.

5. Followability Score (1–10)

Rate how easy the sentence is to follow for a reader. This is strictly about comprehension — can someone reading this understand it without difficulty? It has nothing to do with style, tone, rhetoric, or whether the content is insulting. A long run-on sentence that flows naturally and is easy to follow should score high. A short sentence with garbled syntax should score low. You MUST provide a brief explanation for your score.

6. Fact-Check (if applicable)

  • Using the decomposed claims from step 3, fact-check every factual claim individually.
  • Do a quick verification (web search if needed) — do not run a full fact-check workflow; just confirm or deny the claim with a brief explanation and source.
  • Evidentiary Standard Rule: Do NOT assert a "counter-fact" to debunk the sender unless you are highly certain of it and have concrete proof. If you are uncertain, you cannot mark the sender's claim ❌ False based on a guess. However, if you have partial or suggestive counter-evidence that you are not certain of, mark the claim ⚠️ Uncertain and present the counter-evidence with appropriate hedging (e.g., "some sources suggest X, but this is not confirmed").
  • If the claim checks out, mark it ✅ Confirmed with a one-liner.
  • If it's wrong or misleading, explain why.

7. Credibility Assessment (if applicable)

  • Using the decomposed claims from step 3, assess the credibility of every opinion or prediction individually.
  • Do NOT run the full credibility-check workflow. Just evaluate it briefly: Is it well-grounded? What evidence supports or undermines it? Rate credibility (e.g., low / moderate / high) with a short explanation.

8. Sentence Verdict

After completing the claim-by-claim analysis above, provide a brief holistic assessment of the sentence as a whole: considering all of its facts, opinions, and predictions together, how does the sentence land overall? Is it fundamentally sound, mixed, or off-base? One to two sentences max.

II. Word Usage Question

After completing the sentence-by-sentence breakdown, answer: Did the sender use any word whose meaning they did not understand? List any complex or unusual words and confirm whether they were used correctly.

III. Rhetorical Technique

Assess the rhetorical technique of the email in context:

  • What strategy is the sender using?
  • If there is a history of prior emails, how does this email fit into the arc of that communication?
  • Critical: Do not evaluate the email's rhetoric in isolation. Consider the full context — what the sender has tried before, who the audience is, and what the sender is actually trying to achieve. A confrontational email aimed at deprogramming a sycophantic audience is not the same as a confrontational email aimed at persuading a neutral party. Score accordingly.
  • Tone and Execution: Consider if apparent "sloppiness" (like typos or unpolished sentences) undermines the sender's credibility, or if it instead functions as a deliberate or coincidental signal of contempt/disrespect for the recipient.
  • Attachments/Evidence: If the email includes or references an attachment (e.g., a database of facts or records), evaluate how the existence and thoroughness of that attachment functions as a rhetorical payload, distinct from the email text itself.
  • Argument Architecture: Strip away the tone, the surface vocabulary, and the grammar. What is the underlying skeleton of the argument? Map the sequence of ideas and explain precisely how the sender structurally moves the reader from the opening premise to the final conclusion.

IV. Effectiveness Score (1–10)

Rate how effective the email will be at achieving the sender's apparent purpose. Critical Methodology:

  1. Identify the exact point the sender is trying to make to their specific audience.
  2. Imagine the absolute most persuasive ("10/10") version of an email designed to make that exact point to that exact audience.
  3. Grade the sender's actual email strictly relative to that 10/10 ideal. Do NOT grade the email based on whether you (or the audience) disagree with the core premise. Penalize it only if its tactical execution (rhetoric, framing, insults, logical structure) is worse than the "10/10 ideal version" of making that exact same argument.

V. Output Format and Rules

  • Strict Formatting: The output for section 1 MUST be incredibly concise. Do not write a book. Under each sentence heading, output the breakdown as a simple, un-nested bulleted list.

  • Example format: Sentence 1

    • Original: [text]
    • In my own words: [text]
    • Claims: [Fact: ..., Opinion: ..., Feeling: ...]
    • Grammar/Usage: [notes, if any]
    • Followability: [Score/10] - [explanation]
    • Fact-check: [✅ Confirmed / ❌ False] - [explanation]
    • Credibility: [Low/Moderate/High] - [explanation]
    • Verdict: [holistic sentence assessment]
  • Formatting headings: Use italics for bullet headings (e.g., Original:, In my own words:), not bold. Sentence headers like Sentence 1 should remain bold.

  • Bold ONLY actual issues: Only use bold text for actual errors, problems, or false/misleading claims. Do NOT bold scores (like 9/10), positive confirmations (like ✅ Confirmed), credibility ratings (like High/Moderate), or bullet headings. The goal is to make issues pop out immediately so the reader can scan quickly without being distracted by bolded structural elements.

  • Create a numbered folder per the folder creation policy and save the analysis as analysis.md.

  • Write a README.md in the folder.

VI. Analytical Integrity

Before rendering any fact-check or credibility assessment, and when handling follow-up questions or debates about your analysis, you must read and strictly apply analytical_integrity skill.

The Output

When you run this workflow, the AI agents will generate the following folders and files:

  • README.md — Overview of the email being analyzed and context notes.

    • Example: 001_email_analysis\​README.md
  • analysis.md — The full sentence-by-sentence breakdown with original text, rephrasing, grammar flags, fact-checks, credibility ratings, and followability scores.

    • Example: 001_email_analysis\​analysis.md

How to Set This Up

Option 1: The Easy Way (Automated)

Just point Antigravity to this webpage and ask it to figure it out for you. Antigravity can read this documentation, copy the workflow script, and automatically generate all the required skill files in the correct directories on your machine.

Option 2: The Hard Way (Manual Copy & Paste)

If Antigravity fails to set this up automatically, you will need to manually copy the scripts into your local directories:

  1. Copy the raw workflow script from the "Under the Hood" section above.
  2. Save it as C:\Users\[Your Name]\.gemini\antigravity\global_workflows\email-analysis.md.
  3. You must also click every hyperlinked skill file and save its contents into your skills directory. You must do this for every single skill file linked in the workflow. For example, the legal_analysis_partner skill must be saved to C:\Users\[Your Name]\.gemini\antigravity\skills\legal_analysis_partner\SKILL.md.
  4. Once all files are saved, open Antigravity and type /email-analysis in the chat to run it.