The first dose is free.

You are helping me prepare a skeptical first-read of the attached document. This is the first of four passes. Your job in this pass is descriptive only — extract structure, facts, and the explicit claims the author is making. Do not yet evaluate, judge, or interpret. The skeptical work happens in the next prompt.

## Citation rules

Cite page numbers using the physical PDF page count, where page 1 is the very first page of the file (including any cover, contents, or front-matter page). The document itself may have no printed page numbers — in that case still use physical PDF page numbers.

For sections 3 (Key facts) and 5 (Explicit claims), each item must include both:
- the PDF page number, and
- a short anchor quote of 4–8 words copied verbatim from the document, distinctive enough that I can Ctrl-F to find it.

Format: (p.X — "anchor quote")

For sections 2, 4, and 6, page number alone is sufficient.

## Output

Begin with this page-numbering sanity check as the very first thing in your response:

> Page check. Total PDF pages: N. PDF page 1: [one-line description of what's on it]. PDF page 2: [one-line description]. PDF page 3: [one-line description].

Then produce the following sections in markdown.

## 1. Document overview
One short paragraph: what kind of document is this, who produced it, what period or scope it covers, and what its evident purpose is.

## 2. Structural map
List the document's main sections in order, with one line per section describing what it contains. Identify which sections are narrative vs. tabular vs. legal/standard boilerplate.

## 3. Key facts (with citations)
A table or list of the headline facts, numbers, and dates the document presents. Include comparison values where presented. Each item: PDF page + anchor quote.

## 4. Named entities
People (with roles), organizations, products, places, regulators, and any other proper nouns that recur. Group by category. Brief one-line description of role/relevance.

## 5. Explicit claims the author is making
The substantive things the document asserts — performance claims, strategic claims, forward-looking statements, risk acknowledgments, commitments. One per line. Each item: PDF page + anchor quote. Preserve the author's framing; don't paraphrase.

## 6. Dates and timeline
What period the document covers, what comparison period it uses, and any upcoming dates / milestones / deadlines it cites.

Constraints:
- Be exhaustive on facts, brief on prose. Reference index, not narrative.
- Do not flag concerns, omissions, weasel words, or anything you find suspicious. That's the next prompt's job.
- If a section of the source document doesn't apply, skip it — don't pad.

You've now produced the anatomy. Time for the skeptical pass.

This is pass 2 of 4. Your job is to identify (a) the substantive claims in the document worth testing and (b) what the author appears to be downplaying, omitting, or framing favorably. This is where you apply judgment — but with discipline.

I am reading this document as [YOUR ROLE]. Calibrate the lens accordingly: what would matter to me, what would I miss in a casual read, what would the author prefer I not focus on.

If you do not have the anatomy from pass 1 in your context, ask me to paste it before continuing.

Produce two lists, plus a short coda.

## A. Claims worth testing

The substantive assertions in the document where the evidence is asserted but not fully shown. Look especially for:
- Causal claims ("X was driven by Y")
- Forward-looking statements ("we expect…", "we are confident…")
- Comparative or qualitative claims ("strong", "leading", "favorable")
- Claims about external conditions ("the macro environment is improving")

For each, give:
- The claim, in the author's own words (verbatim quote)
- Page reference (PDF page + anchor quote)
- Why it deserves scrutiny — what's missing, assumed, or unproven, calibrated to my [YOUR ROLE] lens
- What would resolve it: a specific data point to find, a question to ask, or a check to do

Aim for 6–12 items. Pick the sharpest, not the most.

## B. What the author appears to downplay, omit, or frame favorably

Where the document makes a choice that benefits the narrative. Look especially for:
- Big risks mentioned briefly in dense paragraphs or buried in footnotes/appendices
- Numbers framed selectively (rate without absolute, period-over-period without trend, total without breakdown)
- Restated facts, methodology changes, or shifts in scope/comparison base
- Things conspicuously absent (information, evidence, or comparisons the document type would normally include but doesn't)
- Asymmetric terms in contracts/MOUs (obligations heavier on one party, rights heavier on the other; termination, indemnification, IP, audit, confidentiality, dispute resolution)
- Weasel words that soften specific commitments
- Tone shifts between sections (assertive narrative vs. cautious technical sections, footnotes, or appendices)

For each, give:
- What the author said or didn't say
- Page reference (PDF page + anchor quote where applicable)
- Why this matters from my [YOUR ROLE] lens
- What it would look like if presented straightforwardly

Aim for 6–12 items.

## C. Tone and pattern coda

A 3–5 sentence assessment of: where the document is strongest in transparency, where it leans most into spin, and any pattern that ties multiple items together.

Constraints:
- Be specific and evidence-based. Every flag must point to a concrete passage.
- Don't manufacture issues that aren't there. If a section is straightforwardly disclosed, leave it alone.
- Distinguish concerns about the business/counterparty from concerns about the document's framing. Both matter; they are different.
- Be concise. Tight 1–3 pages, not exhaustive.

Pass 3 of 4. Take the items from the skeptical pass and turn them into the questions I would actually ask the author of this document if I had 15 minutes with them, using my same reading lens from pass 2.

Calibrate question style to the role: a board member asks to understand and preserve the working relationship; a journalist asks to find a story; a credit analyst asks for specific data; a buyer asks to protect downside; a fundraising director asks to clarify obligations before signing.

## Top 6–10 questions, grouped by theme

Group questions into 3–5 themes so the conversation flows naturally — don't ricochet between unrelated topics.

For each question:

**Q[n]. [The question — plain language, answerable in 60–90 seconds]**
- Why I'm asking: [the underlying flag from the skeptical pass this consolidates]
- A good answer would: [what reassurance looks like — specific, not vague]
- A concerning answer would: [what should escalate — vagueness, deflection, missing data, or a substantive negative]
- Follow-up to pursue: [data to ask the author for if reachable, sources to verify against, or research to do independently]

## Coda — what to listen for in the room

3–5 sentences on behavioral cues or tells worth watching for given my role, beyond the content of answers.

Constraints:
- Pick the questions that, if answered, would change my assessment most.
- Consolidate ruthlessly: where two skeptical-pass items point to the same underlying issue, ask one question.
- Use plain language, not analyst jargon.
- Phrase questions so a reasonable person can answer in 60–90 seconds.
- Where the underlying issue is sensitive, frame to invite explanation, not to embarrass.

Final pass. Take everything from passes 1–3 and assemble a one-page first-read brief I can use as a standalone artifact — share with colleagues, take into a meeting, file as my read of this document.

The brief must stand on its own without reference to the earlier outputs.

## Output format — rendered markdown, not a code block

Produce the brief as rendered markdown. Do NOT wrap it in a code block. I will select the rendered text manually and copy it; my destination (Word, Gmail, Google Docs) will render the formatting via the rich-text clipboard. Wrapping in a code block defeats this — code-block content copies as plain text only and markdown syntax shows up literally on paste.

If I later ask for the raw markdown source (e.g., for Notion or another markdown-aware tool), provide it then inside a code block. Default is rendered.

## Brief structure (as clean rendered markdown)

1. H1 heading: `# First-read brief: [Document title]`
2. Metadata block (four lines, each "**Label:**" followed by value):
   - **Read for:** [persona/role from earlier passes]
   - **Document type:** [from anatomy]
   - **Period or scope:** [from anatomy]
   - **Read date:** [today's date]
3. H2 `## TL;DR` followed by three sentences as a single paragraph. (1) What this document is and what it claims. (2) The most important caveat or counter-narrative. (3) The single thing the reader most needs to act on or ask about.
4. H2 `## Headline numbers / facts` followed by 3–5 bullets — the bare minimum a reader needs. Page references where useful.
5. H2 `## What's worth testing` followed by 3–5 items. Each: one sentence stating the claim or framing + one sentence on why it deserves scrutiny. Page reference + anchor quote for each.
6. H2 `## What the document downplays or omits` followed by 3–5 items. Each: one sentence on what's quiet or absent + one sentence on why it matters from my role's lens. Page reference + anchor quote where applicable.
7. H2 `## Top questions to pursue` followed by 3–5 items. Each: the question in plain language + one phrase on what kind of answer I'm watching for.
8. H2 `## Bottom line` followed by two sentences. The pattern that ties this read together.

## After the brief, add a short continuation menu

> Want me to:
> - tighten the brief (3 items per section instead of 5)
> - re-run with a different reading lens
> - go deeper on any single item — give me the bullet number
> - give me the raw markdown source for pasting into Notion or another markdown-aware tool
> - produce a plain-text version for email

## Constraints

- One-page artifact. Be ruthless about cuts.
- One sentence per item is the target. Two only when the second sentence does real work.
- Plain language, no analyst jargon.
- Preserve page + anchor-quote references for the testing and downplays items.
- "Read for" line must say which lens this brief was produced through.
- If you find yourself wanting to add a section or extend a list to "be thorough", resist. Thoroughness lives in passes 1–3; the brief is the prioritization step.

I've uploaded a document. Please run a 4-step skeptical review on it end-to-end. My reading lens is [YOUR ROLE].

Produce the following four passes in sequence in your response, each clearly labeled:

PASS 1 — Anatomy: Start with a page-numbering sanity check, then six sections (overview, structural map, key facts, named entities, explicit claims, dates). Descriptive only. Citations: PDF page + anchor quote for facts and claims, page alone elsewhere.

PASS 2 — Skeptical pass: Two lists. (A) 6–12 claims worth testing — each with verbatim quote, page+anchor citation, why it deserves scrutiny calibrated to my lens, and what would resolve it. (B) 6–12 downplays/omissions/framings — each with what was said or omitted, page+anchor citation where applicable, why it matters from my lens, and what straightforward looks like. Plus a 3–5 sentence tone-and-pattern coda.

PASS 3 — Questions for the author: 6–10 plain-language questions, grouped into 3–5 themes. Each with why I'm asking, what a good answer looks like, what a concerning answer looks like, and follow-up to pursue. Plus a coda on what to listen for in the room.

PASS 4 — One-page brief: Header (read-for lens, document type, period, read date), TL;DR (three sentences), headline numbers (3–5 bullets), what's worth testing (3–5 items with page+anchor), what the document downplays (3–5 items with page+anchor), top questions (3–5 items), bottom line (two sentences).

Citation discipline applies throughout: PDF page + verbatim anchor quote where required. No marketing-speak in your own voice. Every flag points to a passage.