Finance Tools (Financial Analysis)

Reviewed 2026-10-13dataPrompt / Rule# finance # dcf # valuation # excel # financial-modeling

Anthropic's financial-analysis plugin: 13 skills and 7 slash commands for DCF, comps, LBO and 3-statement models in Excel, spreadsheet audits and pitch-deck checks.

Security Reviewed

Low risk: instruction-heavy skills with 5 offline helper scripts; external data only through MCP providers you connect yourself.

Install Finance Tools (Financial Analysis) in Google Antigravity

Antigravity loads skills from .agents/skills/ in your project or ~/.gemini/config/skills/ for all projects (official docs). Full guide

EasiestOption 1: paste this to your Antigravity agent

Install the "Finance Tools (Financial Analysis)" skill for Google Antigravity.
Follow the steps at https://www.antigravityskills.org/install/finance-tools.md (source: https://github.com/anthropics/financial-services/tree/main/plugins/vertical-plugins/financial-analysis).
Before running anything, show me what you will download and whether it includes scripts, and wait for my OK.
Install it into this project (.agents/skills/) unless I say global. When done, check that Antigravity lists it as an installed skill.

The agent reads the install file, shows you what it will download and waits for your OK.

Option 2: install it yourself

1. Clone the source repository

git clone --depth 1 https://github.com/anthropics/financial-services /tmp/fsi

Clone to a temporary folder.

2. Copy the skills into this project
```
mkdir -p .agents/skills && cp -r /tmp/fsi/plugins/vertical-plugins/financial-analysis/skills/* .agents/skills/
```
Copies all 13 skill folders. Copy them together: skills load each other (commands/dcf.md loads comps-analysis before dcf-model). Skip skill-creator if you already have one.

3. Or copy them for all projects

mkdir -p ~/.gemini/config/skills && cp -r /tmp/fsi/plugins/vertical-plugins/financial-analysis/skills/* ~/.gemini/config/skills/

Antigravity reads global skills from ~/.gemini/config/skills/<folder>/.

4. Install openpyxl for .xlsx output
```
pip install openpyxl
```
dcf-model/requirements.txt lists openpyxl (and requests, which no reviewed script imports).
5. Check that Antigravity sees them
Start a new session and ask "What skills do you have?" (Antigravity CLI: /skills). dcf-model, comps-analysis, lbo-model and the others should be listed.

The README does not mention Antigravity; this is a manual copy of the plugin's skills folder. The 7 slash commands are Claude Code plugin commands and are not copied, and the data connectors in .mcp.json need your own subscriptions and separate setup.

Using Claude Code, Cursor or Codex? See other assistants below.

What is Finance Tools (Financial Analysis)?

financial-analysis is the core plugin of Anthropic's Claude for Financial Services repository (anthropics/financial-services; plugin.json author: Anthropic FSI, version 0.1.1). It is for analysts, bankers and finance students who want an agent to build and check valuation models in Excel rather than describe them. We counted the plugin folder at the reviewed commit: 40 files, about 264 KB, with 13 skills (3-statement-model, audit-xls, clean-data-xls, competitive-analysis, comps-analysis, dcf-model, deck-refresh, ib-check-deck, lbo-model, ppt-template-creator, pptx-author, skill-creator and xlsx-author), 7 slash commands (/dcf, /comps, /lbo, /3-statement-model, /competitive-analysis, /debug-model, /ppt-template) and 5 Python helper scripts. The largest file is dcf-model/SKILL.md at 49,630 bytes. The skills build live-formula workbooks (Office JS inside Excel, or Python with openpyxl for a standalone .xlsx) and confirm inputs with you step by step.

How to use Finance Tools (Financial Analysis)?

Install the plugin, then call a command or describe the job. /dcf takes a company name or ticker, /comps the same, /lbo a company or deal, and /debug-model a path to an .xlsx file (argument hints in commands/). The DCF skill walks through inputs, projections, WACC and the equity bridge with you before it builds three sensitivity tables of 25 cells each (75 formula cells, per dcf-model/SKILL.md), and it requires derived cells to be Excel formulas, not pasted values. Data comes from what you give it and from any financial-data MCP servers you connect; comps-analysis tells the agent to use S&P Kensho, FactSet or Daloopa MCPs when available and never web search as the primary source.

Key features of Finance Tools (Financial Analysis)

✓DCF models with WACC, terminal value and three 25-cell sensitivity tables built from live formulas
✓Comparable company analysis with operating metrics, trading multiples and summary statistics
✓LBO and 3-statement model building, with formula and formatting references for the 3-statement skill
✓Spreadsheet audit (audit-xls, /debug-model) for broken formulas, balance-sheet imbalances, hardcodes and circular references
✓Pitch-deck checks (ib-check-deck) with a number-extraction script, plus deck-refresh and PowerPoint template skills
✓validate_dcf.py checks a finished DCF workbook with openpyxl and can write a JSON report

When to use Finance Tools (Financial Analysis)

1Command example (commands/dcf.md argument hint "[company name or ticker]"): "/dcf Microsoft". The agent loads dcf-model and comps-analysis, asks you to confirm data and assumptions stage by stage, and builds the workbook with sensitivity tables.
2Our example: "/debug-model ./models/target_lbo.xlsx". audit-xls checks the workbook for broken formulas, balance-sheet and cash tie-outs, hardcoded overrides and circular references, and reports what it found.
3Our example: "Build a trading comps table for these five companies from the numbers in peers.xlsx." Without a data MCP connected, comps-analysis works from your file and should say where each number came from.

Installation & Setup Guide

Agent Rules / Prompt Setup

Follow these steps (taken from the project's own docs) to configure Finance Tools (Financial Analysis) in your development environment.

Step 1: Add the marketplace (Claude Code)Claude Code

claude plugin marketplace add anthropics/financial-services

README, Getting Started, Claude Code.

Step 2: Install the core plugin (Claude Code)Claude Code

claude plugin install financial-analysis@claude-for-financial-services

README, Getting Started, Claude Code: install financial-analysis first ("Core skills + connectors"); named agents and the investment-banking and equity-research bundles are optional.

Step 3: Claude CoworkCowork

README, Getting Started, Cowork: Settings, Plugins, Add plugin, then paste https://github.com/anthropics/financial-services or upload a zip of a folder under plugins/.

Step 4: CursorCursor

Not documented. The README covers only Cowork, Claude Code and Claude Managed Agents, and we found no Cursor instructions in the repository.

Step 5: CodexCodex

Not documented. The README does not mention Codex. The skills are plain SKILL.md folders, but the project gives no Codex path or command, so we do not list one.

Usage Example & Verification

Prompt / Request Example

/dcf [company name or ticker] (the /dcf command and its argument hint from commands/dcf.md)

Installation Confirmation Method

In Claude Code, type / and look for /dcf, /comps and /debug-model after installing the plugin. In Antigravity, ask "What skills do you have?" (Antigravity CLI: /skills): dcf-model, comps-analysis and the other skill names should be listed (the slash commands are not copied by the manual Antigravity steps).

Security Review

Static review

Reviewed Date

2026-10-13

Audited Ref / Commit

574ed3624aebd0418c7e96cd101262f30210ab26 (2026-09-21)

Reviewer

antigravityskills.org (static code review)

Low risk: instruction-heavy skills with 5 offline helper scripts; external data only through MCP providers you connect yourself.

Full review: scope, findings, risks

Audit Scope:plugins/vertical-plugins/financial-analysis/ at commit 574ed36 (also HEAD on 2026-10-13): file list and sizes of all 40 files, .claude-plugin/plugin.json, .mcp.json, hooks/hooks.json, all 7 commands/*.md headers, skills/*/SKILL.md descriptions, dcf-model (SKILL.md, requirements.txt, scripts/validate_dcf.py), comps-analysis/SKILL.md data-source rules, imports of all 5 Python scripts; repository README.md and LICENSE.

Key Findings:Mostly Markdown instructions. The 5 Python scripts (validate_dcf.py, extract_numbers.py, init_skill.py, package_skill.py, quick_validate.py) import only the standard library plus openpyxl or yaml and contain no network, subprocess, eval or exec calls; validate_dcf.py can write a JSON report and package_skill.py writes a zip. hooks.json is empty. .mcp.json declares 12 remote HTTP MCP servers and fails JSON parsing.

Identified Operational Risks & Permissions:Company and deal data may flow to the third-party data providers you connect. Outputs are draft models that need professional review. The malformed .mcp.json may stop connectors loading; recalc.py, referenced by the DCF skill, is missing. Static review only; nothing was run.

Static review on 2026-10-13 of anthropics/financial-services at commit 574ed3624aebd0418c7e96cd101262f30210ab26, plugins/vertical-plugins/financial-analysis (40 files, Apache-2.0). The plugin consists of 13 Markdown skills with reference files, 7 Markdown slash commands, an empty hooks/hooks.json and an .mcp.json listing 12 remote HTTP MCP servers (Daloopa, Morningstar, S&P Global Kensho, FactSet, Moody's, MT Newswires, Aiera, LSEG, PitchBook, Chronograph, Egnyte, Box), which is not valid JSON at this commit. dcf-model/scripts/validate_dcf.py loads a workbook with openpyxl and checks formulas; ib-check-deck/scripts/extract_numbers.py parses deck text locally; the three skill-creator scripts create, validate and zip skill folders. None of them import requests or open network connections, although dcf-model/requirements.txt lists requests. The skills instruct the agent to confirm each modelling step with the user and to write formulas rather than hardcoded values. Static inspection only: we did not install the plugin or run its scripts.

Operational Notes & Known Limitations

Known Limitations

•Not advice: the README says nothing in the repository is investment, legal, tax or accounting advice and every output is a draft for review by a qualified professional.
•Broken connector file (we checked): plugins/vertical-plugins/financial-analysis/.mcp.json lists 12 remote MCP servers but is not valid JSON at the reviewed commit (a missing comma after the egnyte entry and a missing closing brace for box), so connector loading may fail until it is fixed. The file is the same at HEAD.
•Missing helper (we checked): dcf-model/SKILL.md tells the agent to run recalc.py before delivering a standalone .xlsx, but no recalc.py exists anywhere in the repository's 363 files at that commit.
•Best data needs paid accounts: the connectors are Daloopa, Morningstar, S&P Global Kensho, FactSet, Moody's, MT Newswires, Aiera, LSEG, PitchBook, Chronograph, Egnyte and Box.
•Documented only for Claude (Cowork, Claude Code, Managed Agents); Antigravity works by copying skill folders, and Cursor and Codex are not covered.

Frequently Asked Questions

Is the Financial Analysis plugin safe?

In our static review the 5 Python scripts use only local libraries (standard library, openpyxl, yaml) and make no network calls; hooks.json is empty. External data only flows through the MCP connectors you choose to sign in to.

How do I install it in Antigravity?

Clone anthropics/financial-services and copy everything in plugins/vertical-plugins/financial-analysis/skills/ into .agents/skills/ or ~/.gemini/config/skills/, then ask "What skills do you have?" (Antigravity CLI: /skills).