Feat/cf dev cutover

[breardon2011]

No description provided.

[mintlify]

Preview deployment for your docs. Learn more about Mintlify Previews.

Project	Status	Preview	Updated (UTC)
opencomputer	🟢 Ready	View Preview	May 7, 2026, 9:17 PM

💡 Tip: Enable Workflows to automatically generate PRs for you.

[2027-evals]

✅ Eval complete for commit ce98ab3

URL	Mapping
digger.dev	→ opensandbox-feat-cf-dev-cutover.mintlify.app
opencomputer.dev	→ opensandbox-feat-cf-dev-cutover.mintlify.app

Ran evals with prompts:

📉 -10.8 pts · complete the getting started guide at https://opencomputer.d — C (65.4/100) from B (76.2) · View metrics

Prompt text:

complete the getting started guide at https://opencomputer.dev

Verdict:

OpenComputer has solid documentation and a well-typed SDK, but an AI-opaque SPA homepage and missing redirects create significant discovery friction that inflates agent tool-call counts well beyond what the task requires.

Friction points:

• 🔴 Main marketing site (opencomputer — .dev) is a JavaScript SPA with no server-rendered content — AI agents and crawlers receive only a one-line title, making the docs completely undiscoverable from the primary domain.
• 🔴 The URL path /getting-started returns a 404; the actual quickstart lives at /quickstart with no redirect, causing agent detours and wasted tool calls — .
• 🔴 The quickstart code example uses sandbox — .commands.run() which is marked @deprecated in the TypeScript types in favor of sandbox.exec — the guide should use the current API surface.
• 🟡 No OpenAPI / Swagger spec is published, limiting automated client generation and making it harder for AI agents to reason about the full REST API surface — .
• 🟡 The docs are not indexed on Context7, reducing the SDK's presence in AI training data and retrieval-augmented documentation toolchains — .

Result: C (65.4/100)

delta vs baseline: -10.8 pts

Dimension	Baseline	This PR
Setup Friction	86	86
Speed	75	67
Efficiency	54	23
Error Recovery	100	93
Doc Quality	70	60

Stats: 3m 40s · 31 tool calls · 1 error · 1 interruption · $1.80

View report → · View trace →

---

Evaluating agent experience using 2027.dev · View dashboard