# Ivy Health Dashboard

A small operational-evidence reporting product: distinguish a recent observation, an expired observation, a historical event and missing evidence. A successful process exit, a backup file and a tested recovery are different claims.

This is a curated example assessed September 23, 2026, not a live monitor. Historical entries summarize reported operational events; underlying private receipts are omitted. The capacity entry demonstrates an illustrative seven-day freshness rule. The package proves rule behavior and reproducibility, not the truth of input assertions or current service health.

## Build and inspect

Python 3.11+; no external dependencies or network for the build. From the extracted source directory:

```sh
python3 tools/public_health_dashboard.py --input tests/fixtures/ivy_public_dashboard_evidence.json --output-dir out
python3 -m http.server 8000 --bind 127.0.0.1 --directory out
```

Open localhost:8000. The build creates index.html, status.json, evidence.html, README.md and a reproducible source.zip. Keep the package files together: the builder includes its source, tests, input and README in the download. For tests, install pytest in a virtual environment, then run:

```sh
python3 -m pytest -q tests/test_public_health_dashboard.py
```

The downloadable package includes the renderer, curated input, adversarial tests and this README. The product download is [source.zip](source.zip). The [method page](evidence.html) explains the input and recovery boundaries; [status.json](status.json) preserves assessment time, observed time, applied freshness window and age.

## How it works

A person curates the input and determines what its evidence establishes. Python checks fields/types/dates, rejects future observations and inconsistent historical/unknown windows, calculates exact age, and escapes HTML. Current means at or within the supplied freshness window, not healthy. Unknown stays unknown. The text denylist is only a tripwire; human source review and removal of private information are still required.

Ivy Control is the broader private operational system; it is not included. This package has no collectors, credentials, service connection, private receipts, schedulers, alerts or deployment configuration. It does not demonstrate a whole system restore, recurring backups or live reliability.

Buddy built this reporting surface with AI-assisted engineering and review. Product decisions and publication remain human responsibilities. The owned renderer, tests, documentation and curated example in this standalone package are licensed under the MIT License; see LICENSE. The grant does not cover private receipts, third-party material or the broader operational system.