systems registry

Infrastructure surfaces
for evidence-aware
AI observability.

This registry describes SIGNALWATCH architecture boundaries: what is ingested, what is observed at runtime, what is derived, and what must remain unavailable until real data exists.

system topology
conceptual map
runtimesourcesmodel outputevidenceboundary
real source

Claims need source records or ingestion events.

model output only

Confidence and detections come from inference.

unavailable is valid

Missing data stays visible instead of guessed.

no prefilled claims

Protocols do not ship with conclusions.

runtime telemetry
instrumented

Websocket state, collector health, reconnect behavior, and runtime status are surfaced as operational state, not ecosystem facts.

trace / heartbeat
trace / collector status
trace / retry counters
source ingestion
provenance-aware

Research, policy, release, forum, and repository sources are normalized with timestamps and source references attached.

trace / source url
trace / published timestamp
trace / collector route
perception outputs
browser-derived

Detections, confidence, empty frames, and continuity markers come from browser-side model output history.

trace / COCO-SSD
trace / frame timestamps
trace / class continuity
claim boundary
enforced

The system separates facts with sources, summaries, conceptual demos, and unavailable states.

trace / no fabricated detections
trace / no prefilled metrics
trace / unavailable states
technical substrate
implementation registry
frontend

Next.js / TypeScript / Tailwind / Framer Motion

backend

FastAPI / WebSockets / provenance-aware storage

runtime

Docker-ready services / Railway deployment path

evaluation

browser-side model runs / evidence packet export shape

systems evidence boundary
real inputs / real outputs / explicit unavailable states
source data

papers, releases, policy pages, ingested records

traceable
model behavior

browser-side detections, confidence, timing, empty frames

observed
evidence packet

timestamps, frame history, continuity markers, exported JSON

recorded
claim boundary

missing data stays unavailable; outcomes are not prewritten

enforced
real source

Claims need source records or ingestion events.

model output only

Confidence and detections come from inference.

unavailable is valid

Missing data stays visible instead of guessed.

no prefilled claims

Protocols do not ship with conclusions.

operating principles
design constraints
traceability

Operational surfaces should show where information came from and what transformation touched it.

observability

Runtime behavior, collector state, source activity, and model-output gaps stay visible.

robustness

Perception workflows are evaluated under degraded conditions without synthesizing outcomes.

restraint

The interface stays calm and instrumented so evidence is easier to inspect.