SecuritySmith · Implementation handoff

Explorer roadmap

Four network-observability relationships. Real data where deployed, native ThousandEyes tests where needed, and permanent historical replay. Status baseline: October 31, 2026. This is a plan—not an indication that uncreated tests are live.

Measurement relationships

1Cloud Agents → Website

TEST NOT YET CREATED

Native HTTP Server test targeting https://securitysmithrpc.com/. Reuse five selected geographic Cloud Agent perspectives (verify each agent in the test editor). Website delivery uses Hostinger CDN, so these measurements observe the CDN-served application path, not necessarily the origin. The Frankfurt VPS is accessed directly and remains outside the CDN.

  1. Create HTTP Server test, 5-minute interval; enable network measurements and path visualization.
  2. Verify real HTTP and network results in ThousandEyes.
  3. Add returned test ID to private collector and strict exporter allowlist; expose #1 as real measured data.

2Cloud Agents → Frankfurt VPS 🇩🇪

OPERATIONAL

Existing HTTP Server test 562949953717554 targets c3.securitysmithrpc.com/healthz. Five Cloud Agents; direct-to-VPS endpoint outside the CDN; recorded HTTP/network/path-vis results.

Geographic comparison, HTTP timing breakdown, observations, matched-round trends, live feed, and historical replay are working.

3Frankfurt Enterprise Agent → Website

OPERATIONAL

Existing HTTP Server test 562949953717553 uses the installed Frankfurt Enterprise Agent to measure securitysmithrpc.com. Recorded HTTP/network/path-vis results and historical replay.

4Enterprise Agent ↔ Cloud Agent

TEST NOT YET CREATED

Native Agent-to-Agent test: Frankfurt as source, one Cloud Agent as target; enable both directions and use native Path Visualization to examine forward/reverse Internet paths. This is directional network telemetry—not an HTTP Server test.

  1. Confirm target Cloud Agent, TCP/UDP choice, firewall/NAT behavior, and 5-minute interval.
  2. Verify latency, loss, jitter, and native Path Visualization using the source → target, target → source, and both direction controls. Throughput isn't supported for Agent-to-Agent tests involving Cloud Agents.
  3. Build a separate Agent-to-Agent collector/normalizer mapping, whitelist direction-specific public fields, and render #4.

Implementation sequence

StageDeliverableAcceptance criterion
Baseline completeHTTPS Frankfurt VPS, Enterprise Agent, two HTTP tests, private collector, replay normalizer, strict public exporter, Explorer.Both 15-minute systemd timers execute; /thousandeyes/ shows real records and replay.
Next: #1Cloud → Website native test; collector/export integration.New genuine test ID; independent HTTP records visible under #1 with original timestamps.
Next: #4Bidirectional native Agent-to-Agent measurement.Direction-specific measured metrics validated against native ThousandEyes UI; distinct directional archive schema.
HardeningAgent-specific deep links to native ThousandEyes views; feed freshness diagnostics; backup cadence and trial-expiry rehearsal.No credentials, source IPs, or raw hops in public JSON; replay functional after trial stops.
Optional extensionNative widgets when permitted; MCP integration only if it offers distinct value; multi-day analysis.Clearly sourced insights, not a reimplementation of native ThousandEyes.

Architecture diagrams

1. Measurement topology

Four numbered relationships, two existing measurements and two explicitly planned.

Measurement topology connecting Cloud Agents, Frankfurt Enterprise Agent, website and agent-to-agent target.
Open full-resolution SVG

2. Collection and replay pipeline

Private raw archive and normalization separated from curated, browser-accessible output.

Data pipeline from ThousandEyes API through private collection, normalization, public export and browser replay.
Open full-resolution SVG

Operating principles

Use the native ThousandEyes interface for deep diagnostics, native widgets if embedding is supported, and the API only where the Explorer adds distinctive cross-test analysis or durable preservation. No simulated readings, no false live labels. The collector is read-only; new tests will be created separately with appropriately authorized credentials. Confirm trial/usage budget before starting additional recurring tests. The private raw archive is not a public API.

Documentation: ThousandEyes Agent-to-Agent overview · API v7 test creation reference. Independent personal project; not affiliated with Cisco ThousandEyes.