Refactor docs landing-page guardrails section into visual security overview

[Copilot]

This updates the docs landing page (docs/src/content/docs/index.mdx) by replacing the flat security/sandbox prose with a scannable visual section. The Guardrails content is now presented as a short blurb, a flow diagram, and security-layer cards.

• Guardrails section redesign

  • Replaced the long subsection-by-subsection text block with a concise summary of the layered security model.
  • Kept the section anchored under Guardrails Built-In for continuity.

• Security flow visualization

  • Added a Mermaid flowchart to show the end-to-end control path from input through gated write application.
  • Captures layer ordering: read-only token → no secrets in runtime → sandbox/firewall → safe outputs gate → threat detection → scoped write job.

• Layered controls as cards

  • Added a CardGrid with one card per control layer:
    • Read-only token
    • No secrets in agent runtime
    • Sandbox + network firewall
    • Safe outputs gate
    • Threat detection
  • Preserved direct links to deeper docs (architecture, safe-outputs, threat-detection).

## Guardrails Built-In

AI agents can be manipulated by prompt injection, malicious repository content, or compromised tools...
```mermaid
flowchart LR
    INPUT["Repository + Prompt Input"] --> TOKENS["Read-only Token"]
    TOKENS --> SECRETS["No Secrets in Agent"]
    SECRETS --> SANDBOX["Sandbox + Network Firewall"]
    SANDBOX --> SAFE["Safe Outputs Gate"]
    SAFE --> DETECT["Threat Detection Scan"]
    DETECT --> APPLY["Scoped Write Job"]

... ... ... ... ... ```

[Copilot]

Copilot encountered an error and was unable to review this pull request. You can try again by re-requesting a review.