Excel MCP Server
Workflow & Automation · tested 2026-09-01 · by the Hlido desk, not the vendor
In short: MCP server that drives the real Microsoft Excel engine via its COM API — the honest, high-fidelity answer to Excel automation for agents, not another openpyxl file rewriter.
5 PASS · 0 FAIL of 5 public-surface claims
Quick answer
Excel MCP Server scores 80/100 (STEADY) on Hlido’s independent, hands-on test (reviewed 2026-09-01). STEADY (80), medium-high confidence — one of the stronger public surfaces in this cohort. Pricing: Open source (free entry point documented).
Of the MCP servers in this batch, Excel MCP Server has the sharpest and most defensible thesis. Instead of parsing and rewriting .xlsx files directly the way openpyxl-based servers (and Anthropic's own xlsx skill) do — which can silently drop PivotTables, charts and macros and simply cannot execute Power Query, the Data Model or DAX — it automates the actual Excel application through Microsoft's official COM API. Excel itself opens, computes and saves the workbook, so formulas, PivotTables, charts, macros, the Data Model and formatting survive intact, and live operations (refresh Power Query, evaluate DAX, run VBA and Python =PY()) return real computed results. That is a real, correctly-argued technical distinction, not marketing gloss, and the docs make the trade-off explicit rather than hiding it. Maturity signals back it up: a v2.0.5 release, ~645 GitHub stars, an MIT license, a documented surface of 31 tools across ~325 operations, and first-class install paths for VS Code / Copilot / Claude and any MCP client. The honest limits: it is Windows-and-real-Excel-bound by design (the COM API requires a genuine Excel install), so it is not a cross-platform or headless-server play; and because this is a public-surface review, the breadth of those 325 operations is documented but not independently exercised here. Within its lane, this is a well-built, well-scoped, unusually candid tool.
Why STEADY
STEADY (80), medium-high confidence — one of the stronger public surfaces in this cohort. It earns the upper-STEADY band on a genuinely differentiated technical approach (real Excel COM engine vs. file-rewriting), clear maturity (v2.0.5, ~645 stars, MIT), thorough documentation, and refreshingly honest positioning that names its own trade-offs and its competitors' failure modes. Held short of VITAL because the review is a documentation read rather than an execution of all 325 operations, and because the architecture is deliberately Windows/Excel-bound, which caps its addressable surface.
Public-surface checklist
- PASS Homepage loads (required)
- PASS Primary value prop (required) — 'Automate real Excel with AI'
- PASS Cta present (required) — 'Install for VS Code' / All installation options
- PASS Pricing or access — Open-source MIT; free install paths documented
- PASS Evidence or demo — Feature docs + 1-min intro video referenced; 'Show me Excel while you work' live mode
What we saw
4 screenshots captured by the Hlido engine during the reviewed run (run-657a1e3ee88c592d-excelmcpserver-dev). Our own captures — not vendor marketing material.
What it does well
- Automates the genuine Excel COM engine, so PivotTables, charts, macros, the Data Model and formatting are preserved rather than dropped
- Runs live operations agents usually can't reach: Power Query refresh, DAX evaluation, VBA and Python =PY()
- Honest, specific documentation that names exactly how it differs from openpyxl-based servers and the xlsx skill
- Mature and open: v2.0.5, ~645 GitHub stars, MIT license, 31 tools / ~325 documented operations
- First-class install paths for VS Code / GitHub Copilot / Claude and any MCP client, plus a CLI
What it fails at
- Architecturally bound to Windows and a real Excel install (COM API) — not cross-platform, not headless/server-side
- The 325-operation breadth is documented but not independently exercised in this public-surface review
- Requires a licensed Excel, so it's unsuitable for pure open-source or Linux-only automation stacks
Best for
- Agents and analysts who need Excel automation that preserves PivotTables, charts, macros and the Data Model exactly
- Windows users driving Excel from GitHub Copilot, Claude or another MCP client with natural language
- Workflows that depend on Power Query, DAX or VBA — things file-rewriting servers simply cannot do
Not recommended for
- Linux/macOS-only or headless server environments with no real Excel install
- Teams wanting a lightweight, dependency-free .xlsx reader/writer (an openpyxl-based server is simpler for that)
- Users who need a fully cloud-hosted, no-desktop automation path
Pricing & access
- ModelOpen source
- Free entry pointYes — a free tier or open-source edition is documented
- Pricing findable on the public surfacePASS Open-source MIT; free install paths documented (tested 2026-09-01)
Derived from Hlido-held evidence only (engine checklist + editorial text); quotes are verbatim from the scorecard; not vendor-supplied; re-derived daily. Verify current prices on the vendor's pricing page. Last verified 2026-09-01.
Related agents
Agent relevance
API CLI MCP Behavioral-testable
Agentic-Commerce Readiness 75/100 · COMMERCE-READY
Independent readiness for agent delegation & transaction. How it’s scored · check live
Direct MCP server — connect from GitHub Copilot, Claude or any MCP client and drive Excel via natural-language tool calls (31 tools, ~325 operations). Also exposes a CLI. Highly agent-composable within a Windows + real-Excel environment.
Agent-friendly score: 9/10
Score over time
The longitudinal record — every point is the score as published on that date. Raw series.
Evidence
- Automates the real Excel application via its official COM API (Excel opens/computes/saves) — source (2026-09-01) verified
- Preserves PivotTables, charts, macros, Data Model and formatting; runs Power Query, DAX, VBA, Python =PY() — source (2026-09-01) verified
- v2.0.5 release; ~645 GitHub stars; MIT license; 31 tools / ~325 operations — source (2026-09-01) verified



