Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type @ followed by the MCP server name and your instructions, e.g., "@xlsx-audit-mcp Audit this Excel file for circular references and volatile formulas."
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
An MCP server that audits Excel workbooks. Other Excel MCP servers read and write your data — this one reviews your model:
"What feeds the Total cell on the Summary sheet?" — precedent tracing
"If I change this assumption, what breaks?" — dependent tracing, including cells that consume it through ranges like SUM(A1:A40)
"Audit this workbook" — circular references with example chains, volatile functions (INDIRECT, OFFSET, NOW, RAND...), hardcoded constants buried inside formulas, external workbook links, merged cells, extra-long formulas
Spreadsheet mistakes are famously expensive. This is the "trace precedents" discipline auditors apply by hand, exposed to an LLM for a whole workbook at once. Local files only; nothing leaves your machine.
Claude Code
claude mcp add xlsx-audit -- npx -y xlsx-audit-mcpClaude Desktop — add to claude_desktop_config.json:
{
"mcpServers": {
"xlsx-audit": {
"command": "npx",
"args": ["-y", "xlsx-audit-mcp"]
}
}
}Then: "Audit C:\models\budget-2026.xlsx and tell me what looks fragile."
Related MCP server: xlsx-for-ai
Tool | What it does |
| Sheets, dimensions, formula counts, defined names, external links |
| Formulas with addresses and cached values, filterable ( |
| One cell's formula, value, precedents, and dependents (direct + via ranges) |
| Ranked risk report across the whole model |
Reference tokenizer that understands real formulas: string literals are stripped first (the "A1" in INDIRECT("A1") is not a reference), function names can't collide (the G10 in LOG10(...) is not a cell), $ absolutes, quoted sheet names ('My Data'!A1), and ranges are handled.
Shared formulas are materialized. Excel stores filled formulas once with an offset scheme; the loader translates them per-cell (relative refs shifted, absolutes preserved), so dependency queries see what each cell actually computes.
Ranges are never expanded for storage — dependents queries use range-containment tests, and cycle detection caps range fan-out (a SUM(A:A) can't explode the graph; capped ranges are reported, not silently dropped).
No formula evaluation. Cached values from the file are shown instead — no spreadsheet engine dependency.
Known limitations: R1C1 notation and structured table references ([@Column]) are counted but not resolved into the graph.
npm install
npm test # offline tests — synthetic workbooks built in-suite
npm run build # tsc → dist/
node scripts/smoke.mjs # end-to-end: generates a workbook, drives the server over stdioArchitecture: src/xlsx.ts (zip + XML → workbook model) and src/formulas.ts (tokenizer, graph, smells) are pure logic; src/index.ts is the MCP wiring.
MIT
Risk report across the whole model: circular references (with example chains), volatile functions (INDIRECT/OFFSET/NOW/RAND...), hardcoded constants inside formulas, external workbook links, merged cells, very long formulas.
Inputs: path (required).
Formulas with addresses and cached values. Filter with contains (e.g. 'INDIRECT', 'VLOOKUP') and/or sheet.
Inputs: path (required), limit, sheet, contains.
Audit one cell: its formula and cached value, precedents (cells and ranges it reads), and dependents (formulas that read it, directly or via a range). Address like 'Summary!B14'; unqualified addresses use the first sheet.
Inputs: cell (required), path (required), max_dependents.
Structure of an .xlsx workbook: sheets with dimensions and formula counts, defined names, external workbook links, merged cells. The map you need before tracing or auditing.
Inputs: path (required).
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
Scored across 4 tools
Each tool targets a distinct audit aspect: structure, formula listing, single-cell tracing, and overall risk. No overlap or ambiguity, as even the two workbook-level tools (overview vs. audit) have clearly different outputs.
Most tools follow verb_noun pattern (list_formulas, trace_cell, audit_workbook), but workbook_overview is noun_noun, breaking the pattern slightly. The style is otherwise consistent and readable.
Four tools is ideal for an Excel audit server, covering the essential workflow without redundancy or bloat. Each tool has a clear, non-overlapping purpose.
The tool set covers the full audit lifecycle: structure overview, formula discovery, cell-level tracing with precedents/dependents, and a comprehensive risk report. There are no obvious missing operations for the stated domain.
Let AI agents read, check and edit Excel models. Open a workbook or drop in one MCP config. Either way: typed cells, a formula dependency graph, and a verification certificate. Deterministic — no OCR, no model guessing at your numbers.
Your audit methodology inside Claude: findings, risks, controls and workpapers in your team format.
Upload spreadsheets, get plain-English analysis with real computed charts and quality scores.
Validate French FEC accounting files and flag audit risks before a tax audit.
