Autonomous implementation, built on a safety-first foundation
Letting AI touch your live site only works if you can trust the guardrails around it. Here's exactly how Scanedly keeps you in control, your data protected, and every change reversible.
BYOK key handling
Your OpenAI, Anthropic, and Google API keys are encrypted at rest with AES-256-GCM and decrypted only in-memory at call time. Keys are never logged, never shared across workspaces, and never used to train any model.
SSRF-guarded crawler
The crawler validates and constrains every outbound request to prevent server-side request forgery — no reaching internal networks, cloud metadata endpoints, or unintended hosts.
Approval controls
No autonomous change bypasses human review. Every recommendation passes tenant policy validation and environment checks before it's even eligible for approval — and nothing deploys without your explicit sign-off.
Full audit logging
Every crawl, finding, recommendation, approval, and deployment is logged with who, what, and when — giving you a complete, queryable history for compliance and internal review.
Reversible by design
Every implemented change stores a previous-state snapshot before it ships. If a fix doesn't perform as expected, roll it back in a single action.
Least-privilege integrations
GitHub, WordPress, Webflow, GA4, and Search Console connections request the minimum scopes required for their function — nothing more.
Every change passes five checks before it can ship
No autonomous change bypasses this pipeline — regardless of plan or implementation mode.
Tenant policy validation
Environment validation
Artifact generation
Pre-deploy validation
Audit logging
Medium- and high-risk changes are additionally required to ship via PR or staging first — never a direct production edit.
Your data, your models, your control
Scanedly is built BYOK-first: every AI director agent call runs on a key you provide and control. We don't operate a shared model fine-tuned on customer data, and we don't sell or share site data with third parties.
- Provider keys encrypted with AES-256-GCM, per workspace
- No training on your site content or AI outputs
- Crawl data scoped to your workspace only
- Export or delete your data on request
Infrastructure runs on Render with Supabase Postgres and Redis — isolated per environment, encrypted in transit and at rest.
Questions about our security model?
Talk to our team about your specific compliance, data residency, or approval-workflow requirements.
No credit card required · Free plan available · Bring your own AI keys