|
| 1 | +# Controller Case File Template |
| 2 | + |
| 3 | +Use this template for redacted controller-sovereignty evidence. Repository artifacts should contain summaries, not private device records. |
| 4 | + |
| 5 | +## Summary |
| 6 | + |
| 7 | +- Case ID: `<case-id>` |
| 8 | +- Date range: `<start> to <end>` |
| 9 | +- Machine class: `<redacted>` |
| 10 | +- OS family/build: `<OS family and build>` |
| 11 | +- Primary finding: `<one sentence>` |
| 12 | + |
| 13 | +## Controller |
| 14 | + |
| 15 | +- Controller name: `<Spotlight / Photos / APFS / CoreWiFi / Network Extension / updater / other>` |
| 16 | +- Owner: `<first-party | third-party | user | unknown>` |
| 17 | +- Representative services: |
| 18 | + - `<service name>` |
| 19 | +- Resource coalition: `<if known>` |
| 20 | + |
| 21 | +## Trigger model |
| 22 | + |
| 23 | +- Foreground user action observed: `<yes | no | unknown>` |
| 24 | +- Background scheduler observed: `<yes | no | unknown>` |
| 25 | +- Network trigger observed: `<yes | no | unknown>` |
| 26 | +- Filesystem trigger observed: `<yes | no | unknown>` |
| 27 | +- Cloud/sync trigger observed: `<yes | no | unknown>` |
| 28 | + |
| 29 | +## Observed behavior |
| 30 | + |
| 31 | +Describe what the controller did in plain language. |
| 32 | + |
| 33 | +Example: |
| 34 | + |
| 35 | +> Metadata controller compacted index payloads and dirtied several GB of file-backed memory while running non-frontmost background work. |
| 36 | +
|
| 37 | +## Resource impact |
| 38 | + |
| 39 | +- CPU: `<duration / average / peak if available>` |
| 40 | +- Disk writes: `<MB/GB and duration if available>` |
| 41 | +- Memory: `<RSS/footprint if available>` |
| 42 | +- Network: `<interfaces, tunnels, or traffic-class evidence if available>` |
| 43 | +- Power state: `<AC / battery / idle / active>` |
| 44 | +- Platform action: `<none / managed / unknown>` |
| 45 | + |
| 46 | +## Evidence summary |
| 47 | + |
| 48 | +| Evidence | Sanitized value | |
| 49 | +|---|---| |
| 50 | +| Event type | `<disk writes / cpu usage / telemetry loop / network policy / other>` | |
| 51 | +| Command | `<process name>` | |
| 52 | +| Time window | `<redacted or normalized>` | |
| 53 | +| Stack family | `<Spotlight compaction / Photos database rewrite / APFS traversal / Wi-Fi telemetry>` | |
| 54 | +| User activity | `<active / idle / unknown>` | |
| 55 | +| Power source | `<AC / battery / unknown>` | |
| 56 | + |
| 57 | +## User-control gap |
| 58 | + |
| 59 | +- Was the behavior visible in the UI? |
| 60 | +- Could the user defer it? |
| 61 | +- Could the user budget it? |
| 62 | +- Could the user audit what file class, interface, or service was touched? |
| 63 | +- Could the user revise the capability safely? |
| 64 | + |
| 65 | +## Classification |
| 66 | + |
| 67 | +- Evidence confidence: `<low / medium / high>` |
| 68 | +- Resource impact: `<low / medium / high>` |
| 69 | +- User-control gap: `<low / medium / high>` |
| 70 | +- Recommended design response: `<observe / budget / governed profile / redesign>` |
| 71 | + |
| 72 | +## SourceOS design requirement |
| 73 | + |
| 74 | +Translate the case into a requirement. |
| 75 | + |
| 76 | +Example: |
| 77 | + |
| 78 | +> A metadata indexer that exceeds a configured write budget must emit a ControllerBudgetExceeded event with file-class attribution and user-visible policy state. |
0 commit comments