Scheduling
`launchd` install, ensure, reload, and path-aware reinstall flows.
If all you need is one export right now, the upstream project may already be enough. Apple Notes Snapshot earns its keep when you want the workflow to stay observable, schedulable, and easier to recover later.
| Question | Choose upstream only | Choose Apple Notes Snapshot |
|---|---|---|
| Do you only need a one-time export? | Yes | No, you want recurring snapshots |
| Do you care about freshness, failure reason, and logs? | Not much | Yes, operational visibility matters |
| Do you want a local control plane around the exporter? | No | Yes, CLI + optional Web console |
| Do you want a reviewable wrapper around path changes and scheduler setup? | Not necessary | Yes |
| Do you want AI-assisted diagnosis around backup health? | No, deterministic status alone is enough | Yes, an operator next-step assistant would help |
| Do you want MCP-aware agents to inspect local backup state? | No, no agent-facing surface is needed | Yes, a read-only MCP contract is useful |
| Do you want a token-gated same-machine browser/API lane? | No, CLI plus logs are enough | Yes, the Local Web API is useful |
`launchd` install, ensure, reload, and path-aware reinstall flows.
`status`, `verify`, `doctor`, log health, and metrics make the workflow inspectable.
Locks, state files, log rotation, vendor provenance, and token-aware local Web actions.
A calmer local shell around the exporter, so reinstall, path drift, and recovery stay reviewable instead of magical.
Once the operator fit is already obvious, the repo also adds three builder-facing surfaces around the same local facts: AI Diagnose for advisory explanation, a token-gated Local Web API for same-machine browser or HTTP workflows, and a read-only-first MCP surface for agent hosts.
Those layers are real, but they are still second-shelf surfaces. Open Proof if you want the trust boundary first, or For Agents if you already know you need the host-facing contract.