Repository navigation
Simulation and Demo Data
Some HELLO OS surfaces show illustrative values rather than observed activity. A score, hash, agent card, generation count, or snapshot shown on screen is not proof that a live agent ran, a cryptographic operation was verified, or infrastructure was provisioned. The audit dated 20 October 2026 found 14 areas with fake or simulated data.
The audit explicitly called out these examples:
| Surface | What is simulated | Safe interpretation |
|---|---|---|
| Scores and hashes | Random or generated display values | Example UI output; not an evaluation result or verifiable receipt |
| Dashboard agents | Made-up resident agents | Sample/demo residents; not active workers or connected users |
| /evolve | Simulated generation or evolution output | A product simulation; not evidence that a live agent evolved |
| Snapshots | Simulated checkpoint/snapshot records | Illustrative state; not a Firecracker VM or copy-on-write database snapshot |
| Handshake version | Hard-coded 1.0.0 | Stale or placeholder metadata; use the actual runtime/package version after the fix |
| “CI <200ms” | Unsupported speed claim | Not a benchmark until reproducible measurements and test conditions are published |
These are observed examples, not a complete enumeration of all 14 areas. Extend this register when the remaining affected surfaces are identified; do not infer or invent the missing items.
Before synthetic data is shown, identify it with a prominent Simulation or Demo data label near the visual and repeat that distinction in accessible text. Never combine sample records with real user records in a way that makes their origin unclear. Empty states should say when there is no live data.
For generated hashes, do not use words such as “signed,” “verified,” “proof,” or “receipt” unless the value is produced and independently checked by the documented cryptographic path. A random hash-shaped string is only a placeholder.
- Replace random display values with API-backed measurements, or remove the metric.
- Label all sample residents and demo snapshots in the UI and screenshots.
- Make simulated evolution opt-in/demo-only and label its outputs at the source.
- Derive handshake version from the package/runtime version.
- Remove the “CI <200ms” claim until a repeatable benchmark, hardware profile, and test definition exist.
- Add tests that fail if synthetic data is presented as live, plus a review pass over screenshots and copy.