What does this skill do, and when should you use it?
audit-xls audits spreadsheet formulas, data, and common modeling mistakes. It works at selection, sheet, or full-model scope. Full-model audits add balance-sheet, cash-flow, roll-forward, circularity, and reasonableness checks tailored to DCF, LBO, three-statement, merger, and comps models. It reports findings first and does not change the workbook without permission.
It checks for spreadsheet errors such as broken formulas, hardcodes inside formulas, inconsistent neighboring formulas, off-by-one ranges, pasted-over formulas, circular references, broken cross-sheet links, unit mismatches, and hidden rows or tabs. At model scope, it identifies the model type and tests balance-sheet balance, retained-earnings roll-forward, cash tie-out, cash-flow sums, D&A and CapEx matching, working-capital signs, circularity controls, growth and margin reasonableness, terminal-value concentration, and model-specific DCF, LBO, merger, or three-statement issues. It produces a findings table with the sheet, cell or range, severity, category, issue, and suggested fix.
- An investment-banking analyst checks a three-statement model before sending it to a client or investment committee.
- A private-equity team reviews an LBO for debt paydown, PIK interest, management rollover, exit multiple, and fee treatment.
- An equity researcher checks a newly updated earnings model for hardcoded overrides, formula-pattern breaks, and broken links.
- A model owner investigates why a workbook will not balance by quantifying the gap in each period and tracing the break.
- A reviewer performs a DCF quality check covering discount timing, terminal-value discounting, WACC inputs, and tax-shield treatment.
How do you install this skill?
- The skill assumes the host can read and analyze spreadsheets but does not specify required tools, supported formats, or unsupported workbook conditions.
- Audit conclusions may affect financial decisions; human review is mentioned, but evidence retention, sensitive-data protection, and audit logging are not defined.
- No Chinese-language terminology handling or mainland-China network compatibility is documented.
In Claude Cowork, open Settings → Plugins → Add plugin, paste https://github.com/anthropics/financial-services, and select the Earnings Reviewer from the marketplace list. Alternatively, upload a zip of plugins/agent-plugins/earnings-reviewer/. The README does not document a standalone installation command for audit-xls.
tmp="$(mktemp -d)"
git clone --depth 1 https://github.com/anthropics/financial-services.git "$tmp"
mkdir -p ~/.claude/skills
cp -R "$tmp/plugins/agent-plugins/earnings-reviewer/skills/audit-xls" ~/.claude/skills/
rm -rf "$tmp"Generated from the source repository and skill path; it copies only this skill's folder. If the author's install steps above differ, follow those first. To scope it to one project, replace ~/.claude/skills with that project's .claude/skills.
How do you use this skill?
Once installed, send your agent any of these to trigger it:
- audit this sheet,
- check my formulas,
- model won't balance.
With the skill installed and a workbook available, use a trigger such as “audit this sheet,” “check my formulas,” or “model won't balance.” If no scope is supplied, choose selection, sheet, or model; for model scope, identify the model type. The skill reports findings first and does not modify anything without asking.
What are this skill's strengths and limitations?
- Supports progressively deeper audits from a selected range to an integrated financial model.
- Covers balance-sheet, cash-flow, circularity, and model-type-specific integrity checks.
- Uses explicit Critical, Warning, and Info severity levels with suggested fixes.
- Requires balance-sheet gaps to be quantified by period and traced before downstream review.
- The SKILL.md does not specify a spreadsheet platform, file interface, or execution mechanism.
- It reports issues rather than applying fixes automatically, so remediation requires user approval.
- No test suite, accuracy metrics, or platform-validation evidence is provided.
- VBA-driven calculations cannot be fully audited from formulas alone and must be noted.
How does this skill compare with similar options?
Side by side with related skills; every score comes from the same FSRS standard.
| Skill | FS score | Stars | Last updated | License |
|---|---|---|---|---|
| Financial Model Auditor this page ✓ Anthropic · Official | 53 · Use with care | ★ 40k | 1mo ago | Apache-2.0 |
| Model Checker for Financial Audits | 53 · Use with care | ★ 1.8k | 11d ago | Apache-2.0 |
| 3-Statement Financial Model Skill | 51 · Use with care | ★ 1.8k | 11d ago | Apache-2.0 |
| XLSX Spreadsheet Skill (Open Cowork) | 69 · Recommended | ★ 2.2k | 12d ago | MIT |
| Three-Statement Model Builder ✓ Anthropic · Official | 55 · Use with care | ★ 40k | 1mo ago | Apache-2.0 |
How did FollowSkills review this skill?
The skill requires scope confirmation, report-first behavior, and user approval before changes, and it flags macro-driven calculations; however, it does not disclose sensitive financial-data handling, data flows, permission isolation, recovery, or external effects, so points are deducted.
Formula and model-integrity checks are broad, and the skill asks for quantified balance-sheet gaps; however, it provides no reproducible execution procedure, dependency specification, abnormal-input handling, or diagnostic failure messages, so the static score is limited to 8.
The name, target scenarios, three audit scopes, and trigger phrases are fairly clear, with model-type-specific checks; however, non-fit boundaries, environment requirements, Chinese support, and mainland-China reachability are unspecified, so points are deducted.
The document is readable and progressively organized by scope, checks, reporting, and notes; repository context supplies installation guidance, an Apache-2.0 license, and CI governance, but the skill lacks its own versioning, changelog, maintenance owner, dependency notes, examples, and troubleshooting guidance.
The skill defines formula auditing, model-integrity checks, and a structured findings table, making the core task clear; however, no representative output, execution result, or comparative-benefit evidence is provided, so the static score is capped at 7.
The skill text is auditable, and repository workflows validate plugins and scan for secrets; those workflows do not cover the skill's key audit paths, and there is no committed test suite, third-party corroboration, or reproducible execution evidence, so only 4 points are awarded.
Open a dimension to read why it scored that way
Evidence confidence：Low — Mostly static review, author material or a limited demo; useful for discovery, not high-risk decisions.
See the full review method →