Airtable MCP

Why STEADY

STEADY (79) because the tool inventory, licence and install path are all published before install, which is exactly the transparency an agent-consumed component should offer, and the Smithery listing corroborates distribution outside the vendor's own page. Not higher because a 42-tool, fully write-capable server publishes no guidance on token scoping or least privilege — the risk surface is real and entirely undiscussed.

Public-surface checklist

• PASS Homepage loads (required)
• PASS Primary value prop (required) — Let your AI talk to Airtable. 42 tools. Every Airtable operation. Zero config.
• PASS Cta present (required) — Get started →
• PASS Pricing or access — MIT licensed and free; access terms stated on the page
• PASS Evidence or demo — 4 screenshots captured; full tool inventory and runnable config blocks shown on page

What we saw

4 screenshots captured by the Hlido engine during the reviewed run (run-e26c6babf861caca-www-airtablemcp-com). Our own captures — not vendor marketing material.

×

←→

What it does well

• Full 42-tool inventory published by category on the landing page, before install
• One-line npx install with copy-ready config for Claude Desktop (macOS and Windows) and Claude Code
• MIT licence stated plainly on the surface
• Smithery listing gives an independent distribution path and removes token setup
• Covers the genuinely useful hard parts: upsert with deduplication, bulk operations, webhook payload retrieval
• Schema discovery tools let an agent understand a base before acting on it

What it fails at

• No guidance on token scoping or least privilege for a write-capable 42-tool surface
• No discussion of what an agent can damage with base-wide credentials
• Single-maintainer open-source project — continuity is unproven
• No stated versioning or release cadence on the reviewed surface
• Depends entirely on Airtable's API; no statement on behaviour when that API changes

Pricing & access

• ModelOpen source
• Free entry pointYes — a free tier or open-source edition is documented
• Pricing findable on the public surfacePASS MIT licensed and free; access terms stated on the page (tested 2026-08-23)

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 2026-08-23.

Compared to

• n8n direct-agent-access

  n8n is a full workflow platform with an Airtable node among hundreds; this is a single-purpose MCP server exposing Airtable directly to an agent. n8n for orchestrated multi-system flows, Airtable MCP for letting the agent itself hold the connection.

• Zapier raw-tools-vs-managed

  Zapier is trigger-and-action automation with a managed surface and a price; this is free, MIT, and gives the agent the raw tool set. Zapier for non-technical reliability, Airtable MCP for agent-driven work with no per-task cost.

Agent relevance

API CLI MCP Webhook Behavioral-testable

Agentic-Commerce Readiness 82/100 · COMMERCE-READY

Independent readiness for agent delegation & transaction. How it’s scored · check live

Native MCP server, installable with one npx command, with published config for Claude Desktop and Claude Code and a Smithery one-click path. 42 named tools across 11 categories are documented before install, so an agent can determine capability fit without running anything. Webhook tools give an event path in addition to request/response.

Agent-friendly score: 9/10

Evidence

• 42 tools across 11 categories, inventory published on the landing page — source (2026-08-23) verified
• MIT licensed, open source — source (2026-08-23) verified
• Single-command install via npx — source (2026-08-23) verified
• Copy-ready MCP config for Claude Desktop (macOS/Windows) and Claude Code — source (2026-08-23) verified
• Listed on Smithery with one-click install — source (2026-08-23) verified
• Includes write and webhook-creation tools — source (2026-08-23) verified