RXed AI News

AI to the bone.
SCORE 7/10 10/10 criteria · Strong Assessed 2026-09-24 · ARI rubric v1.0

Make — can an agent use it?

Visit make.com
“Visual automation platform for building AI agents, agentic workflows, and integrations without code. 3,500+ app integrations.” — the vendor’s own words

Make is one of the most agent-ready automation platforms: official MCP server, broad API, developer docs with llms.txt and markdown on every page. The main site 403-blocks automated fetches on pricing and legal pages, which is the gap between its developer surface and its marketing surface.

Category
Automation
Agent protocol
Official
Agent payment rails
None published
Site Scan
C (63/100) · crawl-level, 24/09
Tap or hover any criterion to see why it scored that.
FIND & RETRIEVE7.5/10Can an agent get correct information?

Machine-readable surface8/10

Comprehensive llms.txt on developers.make.com indexing 400+ pages across API, Custom Apps, MCP Server, and CLI documentation. All pages available as markdown by appending .md. Sitemap index with 10 sub-sitemaps. However, www.make.com returns 403 on home, pricing, and legal pages to automated fetches — the developer surface is exemplary but the marketing site blocks machines. source

Information retrieval quality5.5/10

Plan tiers (Free/Core/Pro/Teams/Enterprise), operation-based metering, and feature gates are documented in the help center and llms.txt links. However, the pricing page itself (www.make.com/en/pricing) returns 403 to automated fetches, so an agent cannot directly retrieve current prices. Specific operation costs per plan tier require navigating multiple help articles. source

Documentation for machines9/10

developers.make.com hosts full REST API reference with 31 endpoint categories, Custom Apps SDK, MCP Server integration guides for 10+ AI clients, Make CLI docs, and White Label quarterly release notes. All pages in markdown format. Separate llms.txt files on developers.make.com, help.make.com, and academy-content.make.com. Missing: no explicit API changelog or versioning scheme for the core API. source

ACT & TRANSACT6.6/10Can an agent do the job?

Action coverage8.5/10

The API covers CRUD for scenarios (including blueprints and execution logs), connections, webhooks, data stores with record-level operations, teams, organizations, and a dedicated AI Agents endpoint. An agent can create, edit, run, and monitor automation scenarios programmatically. Test mode mirrors production. API access requires a paid plan (Core or above). source

Agent protocol support8.5/10

Official first-party MCP server at mcp.make.com with OAuth and token auth. Documented integrations for Claude Desktop, Claude Code, ChatGPT, Cursor, Windsurf, Gemini CLI, and more. Covers core actions: run scenarios, view/modify scenarios, manage connections, webhooks, and data stores. Not in the official Anthropic MCP registry, but hosted and maintained by Make as a production surface, not a beta. source

Access & auth friction7.5/10

Self-serve API token creation from profile settings with scoped permissions. OAuth 2.0 with PKCE and refresh token flows documented. MCP supports both OAuth (via mcp.make.com) and bearer token auth. OAuth client registration requires requesting access from Make (not fully self-serve). Free tier has no API access at all — the first API call requires a paid plan ($9/mo minimum). source

Agent payment capability1/10

No evidence of any agent payment protocol support (ACP, AP2, MPP, x402). A 'Cashier' API endpoint category exists in the reference but appears administrative. Billing and subscription management are web-UI only. An agent cannot upgrade a plan, purchase credits, or transact on behalf of users. source

Cost & rate fairness7.5/10

Rate limits documented and tiered by plan: Core 60/min, Pro 120/min, Teams 240/min, Enterprise 1,000/min. Queryable via API. No agent-vs-human pricing discrimination — API and MCP consume the same credits as the UI. The free tier excludes API access entirely, but the entry price is low ($9-11/mo). Operation consumption is transparent and trackable. source

TRUST & DURABILITY6.5/10Will it still work next quarter?

Permission & ToS stance6/10

robots.txt allows all user-agents on / with no AI bot blocking. Make's entire business is automation and they publish an official MCP server for AI agents. However, the actual Terms of Service text is unverifiable — www.make.com/en/legal/terms-of-service returns 403. Scoring conservatively: the implicit stance is pro-automation, but the literal terms could not be read. source

Reliability & continuity7/10

Public status page at status.make.com monitoring 10+ components across 6 zones (EU/US, including Celonis zones). Currently 'All Systems Operational'. Incident history shows regular but quickly resolved issues (45-min US2 delays, SSO/login outages). No explicit API versioning or deprecation policy found. Celonis acquisition (2020) added enterprise infrastructure without restricting features. White Label has quarterly release notes, but core API lacks a versioned changelog. source

Sources

Point-in-time assessment — services change terms, prices and APIs often. That volatility is itself scored under Reliability & continuity. Re-verified at least quarterly.