Interactive learning / Technical tools
Explore the numbers behind the systems.
A small, interactive engineering workspace for compression, VPN access models and developer safety. Examples are measured or explicitly labelled conceptual; no live customer systems are connected.
Reproducible sample
Compression: raw vs gzip vs Brotli
Actual byte lengths from generated repetitive text fixtures using Node gzip level 6 and Brotli quality 6. These are not live traffic or universal efficiency benchmarks.
These samples contain extensive repetition and are unusually compressible. For real sites, inspect actual asset sizes, encoding negotiation, cacheability and CPU cost. Compressed images, fonts and archives may not benefit from another compression layer.
Read Cloudflare content-compression guidanceProtocol explorer
VPN, overlay and zero-trust access
Sortable and filterable technology comparisons; no invented Mbps figures or meaningless security scores.
| Technology | Access model | Typical use | Lifecycle | Trade-off |
|---|---|---|---|---|
| WireGuard | Network VPN | Lightweight remote or site links | Modern | Identity, key rotation and access policy need an operating model |
| IKEv2 / IPsec | Network VPN | Site-to-site and managed endpoint fleets | Mature | Cipher negotiation, certificates and routing can be complex |
| OpenVPN | Network VPN | Flexible remote access, varied clients | Mature | Performance depends on implementation; DCO may improve it |
| L2TP/IPsec | Network VPN | Compatibility with older managed clients | Legacy | More encapsulation and less attractive for new deployments |
| Tailscale (WireGuard overlay) | Overlay VPN | Team devices and homelabs | Modern | Control plane and policy choices require review |
Actual VPN throughput depends on CPU, acceleration, packet sizes, MTU, path and configuration. Zero-trust application access is an authorization approach, not an interchangeable tunnel protocol.
Developer workflow
An approval-gated AI agent design
Illustrative stages, not actual autonomous runs or claims of live monitoring.
- 01Observe
Read signals and source state.
- 02Analyze
Check assumptions and confidence.
- 03Approve
Review exact change and rollback.
- 04Execute
Apply only authorized operations.
- 05Verify
Check effects and audit the result.
type AgentAction = { actor: string target: string operation: string approvalId?: string dryRun: boolean}// Example schema; this snippet does not execute anything.