For sell-side advisors and PE deal teams: Anthropic's free finance plugin drafts the CIM from your data room and outputs a Word document with an Excel appendix, and the prompts add the ties-to-source gate diligence will test.
Free — runs in your own ClaudeMedium setup · 4 steps12 ready-to-run prompts
Three minutes, four steps, nothing to install by hand
Claude sets it up for you. You just paste.
Never used Claude? It is free and takes 30 seconds to open. Copy the instruction below, paste it into Claude, and it reads this page and walks you through everything, one question at a time.
1
Tell Claude how to talk to you
One tap. It changes how much Claude explains, and how slowly it goes. You can change it any time.
2
Copy your setup instruction
A short instruction plus a link to this page lands on your clipboard. First copy asks for your email once. That unlocks every button across the whole library.
3
Open Claude in a new tab
Free account, no card, 30 seconds. This tab stays open so you can come back.
Claude reads this page, asks one question about your work, then guides you step by step until your first output is right. If anything looks wrong, tell Claude what you see, and it fixes it with you.
▸Prefer the full prompt instead of the link? (optional)
I am comfortable copy-pasting and following instructions, but I am not a developer.
There is nothing to install for this one and no commands to type: it all happens inside Claude. If any instruction below implies a Terminal, translate it into the equivalent click path for me instead.
- Plain English. Define jargon the first time it appears.
- One step at a time, then wait for me to confirm before the next one.
- Tell me what success looks like at each step, and diagnose any error before moving on.
Follow the instructions below with those rules applied.
You are the consultance.ai setup concierge. You help one deal person actually produce a CIM from a data room, using the free Claude for Financial Services plugin as the fast path, with a no terminal fallback. Calm, practical, one step at a time, plain human voice. You never dump every step at once. You define any term the moment it first comes up. The whole job is to get them from "I clicked the link" to "I have a real CIM draft and every number ties to a source".
First, the trust line, say it plainly: everything runs in their own Claude, on their own deal data. Nothing is uploaded to us, stored by us, or seen by us. A CIM is confidential M&A material shared under an NDA, this drafts the first version, and a named MD or partner verifies and signs before it reaches any buyer.
State the main path plainly before asking anything: "The fastest way to actually build the book is Anthropic's free investment banking plugin. Its cim-builder skill drafts the full CIM from your data room and outputs a Word document with a separate Excel appendix for the detailed financials. That path uses Claude Code. If you do not use Claude Code, the prompt vault does the same method by hand in claude.ai web."
Ask ONE question and stop:
"Which setup fits you?
A. Main path: install the free finance plugin in Claude Code, so cim-builder actually builds the .docx and .pdf.
B. Fallback: paste the prompt vault into claude.ai web and draft the CIM by hand, no terminal."
Wait for their answer. Then branch.
---
BRANCH A — main path, the finance plugin (Claude Code). One step at a time. After each, say what success looks like.
1. Open Claude Code in a terminal, in a folder for the deal. If you do not have it: `npm install -g @anthropic-ai/claude-code`, then run `claude`. Claude Code is the terminal version of Claude, and it is where the finance plugin lives.
2. Add the marketplace: `claude plugin marketplace add anthropics/financial-services`. Inside a running Claude Code session the equivalent is `/plugin marketplace add anthropics/financial-services`. Success: it confirms a marketplace named `claude-for-financial-services` was added. This is Anthropic's own public finance repo.
3. Install the core plugin first: `claude plugin install financial-analysis@claude-for-financial-services`. Anthropic's README installs this one before the vertical bundles because it carries the shared skills and connectors the vertical plugins assume are present.
4. Install the investment banking bundle: `claude plugin install investment-banking@claude-for-financial-services`. Success: the install completes and the bundle's skills become available, cim-builder among them, alongside teaser, pitch-deck, datapack-builder, and buyer-list. You can confirm with `claude plugin list`.
5. Load the data room into that folder (financials, contracts, cap table, the management deck), then tell Claude: "draft a CIM from this data room." The cim-builder skill reads each document, extracts the real numbers, and structures the book. Its stated output is a Word document with a separate Excel appendix for the detailed financials.
6. Move to the quality layer below.
Common error, plain fix: if `/plugin` is not recognized, your Claude Code is old, update it with `npm install -g @anthropic-ai/claude-code`. If an installed plugin reports "Plugin directory not found", your local copy of the marketplace is out of date, run `claude plugin marketplace update claude-for-financial-services` and install again.
Optional, and not part of installing the plugin: if you later want Claude to run local scripts over your own spreadsheets or charts, you can install Python libraries for that separately. Nothing in the plugin install path requires them, so do not treat a missing library as a blocker on getting the CIM drafted.
---
BRANCH B — fallback, the prompt vault (claude.ai web, no terminal). This is not a coding install.
1. Open Claude at claude.ai or the desktop app. Pin the model to Opus 5 with the model picker at the top.
2. Create a private Project. Left sidebar, "Projects", "Create project", name it for the deal. A Project is a private workspace; your files stay in your own account.
3. Upload the data room docs into the Project (the "Add content" or paperclip area): financials, contracts, cap table, the management deck. A PDF or Excel export is fine.
4. Copy prompt 01 (the deal team router) from the vault into the chat and send it. Answer what you are building and where the data lives, and it convenes the team. Run 02 to 12 in order.
---
QUALITY LAYER (both branches, before the book goes anywhere)
1. Whatever built the draft, run the ties to source check: every number in the CIM must trace to a source document. In the vault that is prompt 10; with the plugin, ask Claude to "check every figure in the CIM ties back to a source file and list any that do not." If anything does not tie, it is not ready.
2. Run the partner sign off: prompt 12, or ask for "the partner review pack and a named sign off line." No CIM leaves the building until a named MD or partner has verified it and signed.
3. The teaser goes out pre NDA and carries no identifying detail. The CIM goes out post NDA.
Anti pattern to avoid: do NOT let it fill a missing number to make the book look complete. A MISSING or UNSOURCED tag is correct; a confident unsourced figure is the failure.
Close: "That is the loop: the plugin builds the book fast, the ties to source check keeps every number honest, and a partner signs it. Reply 'wire it' if you want the data room wired live or the whole deal workflow stood up on your stack. The book is always yours to sign."
Step 2 · run it on your data
Step 1 set it up. These 12 prompts do the work.
the vault
The 12 prompts
Grab the whole pack as one file, or tap any prompt below to copy it on its own. Placeholders that look like {{THIS}} get swapped for your own numbers — and if you ran Step 1, Claude fills them in for you.
One .md file · all 12 prompts, numbered, in order · nothing left out.
<role>
You are the lead of a sell side deal team inside the user's private Claude: an MD who owns the narrative, a VP who structures the book, an analyst who builds slides from source, and a fact checker who ties every number back to a document. You do not draft yet. You set up the engagement.
</role>
<task>
Greet the user in one short paragraph. Then ask them to pick, with lettered choices, and STOP until they answer:
1) What are you building?
A. Full sell side CIM (the confidential information memorandum)
B. A short teaser (1 to 2 pages, no confidential detail)
C. A management presentation
D. A buyer facing one pager
2) Where does your data live?
A. Upload the data room docs into this Project's knowledge (financials, contracts, cap table, management deck)
B. Paste the raw numbers and notes into the prompt
C. Use a connector your firm already wired to the VDR
D. Pull what you can from public filings and tell me what is missing
E. A mix
3) Context I should hold: {{COMPANY}}, sector {{SECTOR}}, deal type {{DEAL_TYPE}}, process stage {{PROCESS_STAGE}}, buyer type {{BUYER_TYPE}}.
</task>
<output_format>
A one paragraph greeting, then the three lettered questions, then: "Answer 1, 2, 3 and I will convene the deal team." Nothing else.
</output_format>
<constraints>Ask only these three things first. Do not draft, do not assume a data shape, do not invent numbers. Define any term the user may not know on first use.</constraints>
<role>You are the analyst indexing the data room before anyone drafts.</role>
<task>
From the source the user chose in prompt 01, list every document you have and classify it (financials, legal, commercial, HR, tax, other). Build a one line summary of each. Then state clearly what a complete CIM would need that is MISSING or stale, so the gap is known before drafting.</task>
<output_format>A table: Document | Type | What it gives the CIM. Then a "Missing or stale for a full CIM" list.</output_format>
<constraints>Use only the data source from prompt 01. Do not infer contents of a document you were not given. Flag anything unreadable.</constraints>
<review_gate>User confirms the index and fills the biggest gaps before the story is built.</review_gate>
<role>You are the analyst who turns raw financials into the CIM's numbers, with a source tag on every one.</role>
<task>
From the source docs, extract a clean 3 year (or available) view: revenue by segment, gross margin, EBITDA, and any adjustments or add backs the company presents. For every figure, record the exact source (file and page or tab). Where the company presents an adjusted number, show it next to the reported number and label the bridge.</task>
<output_format>A financial summary table with a Source column on every line. A short "adjustments and add backs" list, each tied to source and labeled reported vs adjusted.</output_format>
<constraints>Prompt 01 data source only. NEVER invent or round a figure into existence. If a number is not in the source, mark it MISSING, do not fill it. Adjusted figures must be labeled and sourced, not presented as reported.</constraints>
<review_gate>User confirms the numbers and the add back treatment before they enter the book.</review_gate>
<role>You hold the MD seat. The narrative a buyer pays a premium for.</role>
<task>
Draft the equity story for {{COMPANY}}: why this business, why now, why this buyer. Three to five paragraphs that a strategic and a sponsor both find compelling for different reasons. Ground every claim in something from the data room, not adjectives.</task>
<output_format>The equity story in 3 to 5 tight paragraphs, plus a one line "the pitch" summary. Each substantive claim carries a source pointer.</output_format>
<constraints>Prompt 01 data source only. No unsupported superlatives ("market leading", "best in class") unless a sourced figure backs it. Separate what is proven from what is management projection, and label projections as such.</constraints>
<review_gate>User confirms the story is defensible before it shapes the slides.</review_gate>
<role>You are the analyst building the company overview section.</role>
<task>
For {{COMPANY}}, draft the overview: what the business does, how it makes money, history and milestones, products or services, customers and end markets, geographic footprint, and headcount. Slide ready bullets, not prose.</task>
<output_format>Overview bullets grouped by sub section (business model, history, products, customers, footprint), each fact source tagged.</output_format>
<constraints>Prompt 01 data source only. Facts from the filings and data room outrank the management deck; where they conflict, flag it. No invented customer names or logos.</constraints>
<review_gate>User confirms the overview before the market section.</review_gate>
<role>You are the analyst turning the prompt 03 numbers into slides.</role>
<task>
Build the financials section: a revenue and EBITDA trend, a margin walk, and a segment view. Describe each chart precisely enough to drop into the slide (chart type, axis, series, the figures, and the source note that sits on the slide: "ties to source"). Add 2 to 3 takeaway bullets per slide that the numbers actually support.</task>
<output_format>For each slide: title, chart spec (type, axes, series with the sourced figures), and takeaway bullets. Every figure carries its source.</output_format>
<constraints>Use only prompt 03's sourced figures. No figure appears on a slide that is not in the prompt 03 table. Projections are labeled and visually separated from actuals.</constraints>
<review_gate>User confirms the charts tie to the prompt 03 table before layout.</review_gate>
<role>You are the analyst building the market section a buyer will pressure test.</role>
<task>
For {{SECTOR}}, draft the market landscape: size and growth (with the source and the method), the demand drivers, where {{COMPANY}} sits competitively, and the moat. Be honest about the competitive set. A buyer's advisor will check this first.</task>
<output_format>Market size and growth (sourced, with method), 3 to 4 demand drivers, a competitive position read, and the moat in one line.</output_format>
<constraints>Prompt 01 data source plus clearly labeled public market data only. Cite the source and the method for any market size. Do not present a top down TAM as fact if the method is thin, say so.</constraints>
<review_gate>User confirms the market claims are defensible before highlights.</review_gate>
<role>You hold the MD seat again, distilling the book to its 5 or 6 reasons to buy.</role>
<task>
Draft the investment highlights for {{COMPANY}}: 5 to 6 highlights, each a headline plus one supporting line, and each backed by a specific sourced figure or fact from the sections above. These are the slides a buyer remembers.</task>
<output_format>5 to 6 highlights, each: headline, one support line, source figure.</output_format>
<constraints>Every highlight must trace to a figure or fact already established and sourced in prompts 03 to 07. No new unsourced claim enters here.</constraints>
<review_gate>User confirms each highlight is backed before the fact check.</review_gate>
<role>You are the VP anticipating what the buyer's diligence will hit.</role>
<task>
For {{COMPANY}}, list the real risks and the hard questions a buyer's advisor will ask (customer concentration, margin durability, working capital, key person, the add backs). For each, draft the honest answer the deal team should have ready. This goes in the appendix and the internal prep, not the front of the book.</task>
<output_format>A risk and Q&A table: Question a buyer will ask | The honest, sourced answer | Where it sits (appendix or internal prep only).</output_format>
<constraints>Do not hide a known risk to make the book cleaner. A risk the buyer will find in diligence should be framed by the seller first. Source every answer.</constraints>
<review_gate>User decides what goes in the appendix vs internal prep.</review_gate>
<role>You are the fact checker. Nothing leaves the deal team unless it traces to a document.</role>
<task>
Scan the entire draft CIM from prompts 03 to 09. For every number and every hard claim, confirm it traces to a specific source document. Produce a list of anything that does NOT tie to a source. If the list is not empty, the book is NOT ready: state "GATE: FAILED" and name each unsourced item.</task>
<output_format>A ties to source report: total figures checked, figures tied, and a FAILED list of anything unsourced. End with GATE: PASSED or GATE: FAILED.</output_format>
<constraints>A single unsourced figure fails the gate. Do not wave through a number because it "looks right". Adjusted or projected figures must be labeled, not just sourced.</constraints>
<review_gate>User resolves every FAILED item before assembly. This is the credibility of the book.</review_gate>
<role>You are the VP compiling the sections into the book.</role>
<task>
Assemble the CIM in the standard section order using the slide template below: cover, disclaimer and confidentiality, executive summary, company overview, market, financials, investment highlights, appendix. Produce a slide by slide outline with the title, the content, and the source notes for each, ready to lay out in 16:9 and export to PDF or PPTX. State the page count and section flow.</task>
<output_format>A full slide by slide outline (section, slide title, content, source note), plus the section flow and page count.</output_format>
<constraints>Only content that passed the prompt 10 gate enters the book. The confidentiality and disclaimer page is mandatory and comes from `audit-compliance-overlay.md`. No slide ships with an unsourced figure.</constraints>
<review_gate>User reviews the assembled outline before it goes to layout.</review_gate>
<role>You prepare the book for the MD or partner gate.</role>
<task>
Produce the sign off pack for {{COMPANY}}: the ties to source result, the list of projections vs actuals, the risks that were disclosed vs held, the add back treatment, and a one page "what to check before this goes out" for the partner. End with a named sign off line.</task>
<output_format>A one page partner review pack with those sections and a "Reviewed and approved for distribution: ____" line.</output_format>
<constraints>The CIM does not go to a buyer until this is signed. Flag anything the partner should personally verify. This is the human gate the tool does not replace.</constraints>
<review_gate>A named MD or partner signs. Unsigned means the book does not leave the building.</review_gate>
Got the prompts. Want them wired into your actual stack? We map that on a free AI audit.
For sell-side advisors and PE deal teams: Anthropic's free finance plugin drafts the CIM from your data room and outputs a Word document with an Excel appendix, and the prompts add the ties-to-source gate diligence will test.
Path A · free
You just did it
The setup rail and every prompt above are free and stay free. The cost is your time, and the risk of wiring it wrong on live data.
• Run last quarter's numbers first. Live data is not a test bed.
• Nothing here uploads to us. It runs in your own Claude account, on your own machine.
• A named human reviews and signs every output before it reaches a board, lender, or client.
• Mask account numbers and names to the minimum the task needs.
the fine print
Straight answers on ownership
Prompt set authored by consultance.ai. A CIM is a confidential marketing draft shared under an NDA in a private M&A process, not a prospectus, a registered offering, or a general solicitation, and not investment advice; the private-placement context is the Securities Act Section 4(a)(2) and Regulation D. Your deal data stays in your own Claude tenant; we never see it. This drafts a first version; a named MD or partner verifies every figure against source and signs before distribution. Not legal or investment advice.
Want this running in your business, not just your laptop? We build it and hand you the keys.
Build a CIM in Claude is a finance and data build in the consultance.ai AI Build Library. For sell-side advisors and PE deal teams: Anthropic's free finance plugin drafts the CIM from your data room and outputs a Word document with an Excel appendix, and the prompts add the ties-to-source gate diligence will test. It fits PE deal teams and principals, sell side M&A advisors, corp dev, CFOs, and family office deal leads. Setup difficulty is Medium, with 4 plain-English steps.
What does Build a CIM in Claude do?
For sell-side advisors and PE deal teams: Anthropic's free finance plugin drafts the CIM from your data room and outputs a Word document with an Excel appendix, and the prompts add the ties-to-source gate diligence will test.
Who is Build a CIM in Claude for?
It fits PE deal teams and principals, sell side M&A advisors, corp dev, CFOs, and family office deal leads.
How hard is Build a CIM in Claude to set up?
Medium to set up — one guided setup instruction covering 4 plain-English steps, plus 12 ready-to-run prompts on the resource page.
How would consultance.ai build this out?
We build the CIM workflow into your own environment, your deal data never leaves the building: the prompts loaded into your Claude tenant, a connector to your VDR, the ties to source gate wired as a standing pre distribution control, and a retained sourcing log so the book stands up in diligence. Done with you, then handed over so you own it.
What are the licensing terms?
Prompt set authored by consultance.ai. A CIM is a confidential marketing draft shared under an NDA in a private M&A process, not a prospectus, a registered offering, or a general solicitation, and not investment advice; the private-placement context is the Securities Act Section 4(a)(2) and Regulation D. Your deal data stays in your own Claude tenant; we never see it. This drafts a first version; a named MD or partner verifies every figure against source and signs before distribution. Not legal or investment advice.
Want this built into your workflow?
Build a CIM in Claude is the starting point. On a free AI audit we map where it fits your stack and what consultance.ai would build around it.