Agent Friendliness Index
The agent era has a new question: can an AI agent actually use this service on your behalf? We score it — 8 criteria, 0–10 each, deterministic mean, every claim sourced and dated. Rubric v0.1. Hover any score for the reason.
| Service | Overall | API surface | MCP / agent protocol | Docs quality | Auth friction | Rate limits & pricing fairness | ToS stance on agents | Machine-readable output | Stability |
|---|---|---|---|---|---|---|---|---|---|
| Stripe Payments | 9.4 Agent-ready benchmark | 10 | 9.5 | 10 | 9 | 9 | 8.5 | 9.5 | 9.5 |
| GitHub Developer platform | 9.2 Agent-ready benchmark | 9.5 | 10 | 9.5 | 9 | 8.5 | 9 | 9 | 9 |
| Cloudflare Infrastructure | 8.9 Agent-ready benchmark | 9 | 9.5 | 9 | 8.5 | 8.5 | 9 | 9 | 8.5 |
| Shopify E-commerce | 7.9 Strong | 8.5 | 8.5 | 8.5 | 7 | 7.5 | 8 | 8.5 | 7 |
| Notion Productivity | 7.8 Strong | 7 | 8.5 | 8 | 8 | 7 | 8 | 8 | 7.5 |
| Slack Communication | 7.4 Strong | 8 | 8 | 8.5 | 6.5 | 6 | 7 | 8.5 | 6.5 |
| X (Twitter) Social platform | 3.9 Agent-hostile | 6 | 2 | 5.5 | 5 | 3 | 3 | 5 | 2 |
| LinkedIn Social platform | 2.7 Agent-hostile | 3 | 1 | 5 | 2 | 3 | 1.5 | 2 | 4 |
HOW WE SCORE
- API surface — Is there a public API covering the service's core actions? Coverage vs what a human can do in the UI.
- MCP / agent protocol — Official MCP server or documented agent integration. Community-maintained MCP counts at half-weight.
- Docs quality — Can an agent or its developer find auth, endpoints and limits without a sales call?
- Auth friction — Self-serve API keys? Agent-compatible OAuth? Or manual review and enterprise gates?
- Rate limits & pricing fairness — Usable limits at accessible tiers; agent traffic priced like human traffic or punished?
- ToS stance on agents — Automation explicitly permitted, silent, or prohibited — including enforcement history.
- Machine-readable output — Structured data (JSON, RSS, schema.org) vs HTML walls, CAPTCHAs and bot-blocking.
- Stability — API and pricing deprecation track record over the last ~24 months.
Scores are point-in-time assessments with full source lists on each service page. Evaluations run on our production research pipeline and are benchmark-checked against an independent frontier-model assessment before publication — quality first, always. Want a service scored? Request it →