RXed AI News

AI to the bone.
SCORE 6/10 10/10 criteria · Capable, with friction Assessed 2026-09-06 · ARI rubric v1.0

Sendcloud — can an agent use it?

Visit sendcloud.com
“Europe's leading shipping platform for e-commerce, connecting 30,000+ merchants to 170+ carriers via a single API.” — the vendor’s own words

A strong API for shipping automation with documented rate limits and multi-carrier coverage, but no agent protocol support and no machine-readable pricing or structured data — an agent can ship parcels but cannot discover or negotiate the service without prior knowledge.

Category
Shipping
Agent protocol
None
Agent payment rails
None published
Site Scan
D (53/100) · crawl-level, 06/09
Tap or hover any criterion to see why it scored that.
FIND & RETRIEVE6.5/10Can an agent get correct information?

Machine-readable surface5/10

Sitemap present and referenced in robots.txt, content renders without JS (1,376 words in raw HTML), and supports markdown content negotiation. However, no llms.txt, no JSON-LD structured data, no schema.org markup, and poor semantic HTML (no H1, no <main>, no <nav>). An agent can read the page but must parse marketing prose to extract facts. source

Information retrieval quality7/10

Pricing page publishes concrete tier prices (€0 free up to 20 parcels, Lite €28/mo + €0.10/label, Growth €87/mo, Premium €175/mo, Pro €639/mo) with feature breakdowns and label caps per tier. Carrier rates shown with examples (DPD from €3.38, GLS from €3.74). Facts are current and specific enough for an agent to answer 'how much does it cost to ship a parcel' — but no machine-readable Offer markup, so extraction requires prose parsing. source

Documentation for machines7.5/10

Developer portal at sendcloud.dev provides API reference for V2 and V3, rate limit documentation, quick-start guides, and client libraries in Ruby, PHP, .NET, and Go. V3 is recommended with clear migration rationale. Documented endpoints include Checkout, Orders, Shipments, Returns, Contracts, Support, and Webhooks. Changelog exists (V2→V3 differences documented). No per-field error taxonomy found, but docs are public and navigable without login. source

ACT & TRANSACT4.8/10Can an agent do the job?

Action coverage8/10

API V3 covers the full shipment lifecycle: create labels, get rates, manage orders with shipping rules, returns management, carrier contract CRUD, tracking webhooks, checkout delivery options, and support ticket creation. 99.9% stated uptime. The API surface closely mirrors what a human can do in the UI — an agent can create a shipment, print a label, and track a parcel end to end. source

Agent protocol support1.5/10

No official MCP server. No .well-known/mcp.json. Not in the MCP official registry. A third-party hosted connector exists via Peliqan (a middleware platform that wraps any API as MCP), but this is a generic wrapper, not a Sendcloud-maintained agent protocol. No WebMCP, no Agent Skills, no API catalog discovery. An agent can only reach Sendcloud through its REST API with prior integration knowledge. source

Access & auth friction6.5/10

API uses Basic Authentication with a Public and Private key pair, obtainable self-serve from the Sendcloud panel settings (Settings → Integrations). Keys are per-integration, not globally scoped, which gives some isolation. No OAuth flow, no granular permission scoping, no agent-compatible token exchange. An agent can authenticate immediately once given credentials, but credentials are all-or-nothing with no read-only or action-restricted variants. source

Agent payment capability1/10

No agent payment protocols (ACP, AP2, MPP, x402). The service charges per label (€0.07–€0.10/label on paid plans) and monthly subscription fees, but payment is handled through a human-managed billing relationship — there is no programmatic way for an agent to pay for label creation or upgrade a plan. Label costs are deducted from a pre-funded balance or invoiced, which is closer to a pre-pay ledger than an agent payment flow, but still requires human setup. source

Cost & rate fairness7/10

Rate limits are documented and reasonable: 1,000 GET requests/min, 100 POST/PATCH/PUT/DELETE requests/min, 15 requests/second burst on unsafe methods. HTTP 429 on exceeding. The free tier (€0/mo, up to 20 parcels/month) is genuinely usable for testing. Paid tiers scale linearly — no evidence of agent-vs-human price discrimination. However, no sandbox environment exists; testing requires using the 'Unstamped Letter' shipping option as a workaround. source

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

Permission & ToS stance5.5/10

Terms (Article 3.1) explicitly prohibit 'scraping, data mining, or any other automated methods to access or extract data from the Services without Sendcloud's prior written consent.' However, the API itself is marketed for automation ('automate the entire shipment lifecycle without manual input'). This creates a split: using the API as designed is encouraged, but any data extraction beyond the API's intended scope requires written consent. Robots.txt is permissive (Allow: /) with no AI-specific rules — policy on crawlers is accidental rather than deliberate. source

Reliability & continuity7.5/10

Public status page at status.sendcloud.com with component-level breakdown (17 Sendcloud components at 100% uptime, 55 integration components at 99.997%, 87 carrier components at 91.2% — carrier reliability is outside Sendcloud's control). API V3 is the current version with a clear upgrade path from V2. The platform serves 30,000+ merchants, indicating production-grade stability. No evidence of sudden API deprecations or pricing overhauls in recent history. 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.