Network activity

A complete list of the network calls MimicScribe makes, what each one carries, and how to inspect them.

Endpoints

Host	When	What’s in it
`mimicscribe.app/api/gemini/...` (Cloudflare)	Cloud meeting summary, talking points, Q&A, transform, dictation refinement	Transcript text and your prompt. Not raw audio.
`mimicscribe.app/api/warm`	Briefly, when a cloud meeting starts	Empty GET — TLS handshake to shave latency off the first real call.
`mimicscribe.app/appcast.xml`	At launch, then every 24h (Sparkle)	GET request. App + macOS version in the User-Agent.
`mimicscribe.app/api/analytics`	App launch and selected events, only if analytics is on	Event name, app version, OS version, anonymous device hash, timestamp.
`mimicscribe.app/api/crash-report`	When a queued crash exists, only if crash reporting is on	Stack trace, app version.
HuggingFace CDN (`huggingface.co` via CloudFront)	When MiniLM and Parakeet models load	Manifest revision check. No user data.

Audio capture, transcription (Parakeet), speaker diarization, echo cancellation, and VAD run on-device. Audio bytes are never sent anywhere.

The Gemini proxy is open source: github.com/mimicscribe/mimicscribe-proxy.

Inspect it yourself

macOS ships nettop (live byte counters per process) and lsof (open sockets):

PID=$(pgrep -x mimicscribe)
nettop -x -L 1 -J bytes_in,bytes_out -p $PID
lsof -i -nP -a -p $PID

For a UI alternative, Little Snitch ($69) prompts on each connection and works on macOS Tahoe. LuLu (free, by Objective-See) is the usual recommendation but currently has a NetworkExtension provider issue on Tahoe 26.2+ — verify it’s actually catching traffic before relying on it.

Local Mode

Local Mode disables every cloud feature for a meeting. The on-device pipeline still runs — transcription, diarization, summary. What changes on the network:

No /api/gemini/... calls. No transcript leaves the Mac during or after the meeting.
No /api/warm prewarm at meeting start.
Analytics and crash reporting stay off if you turned Local Mode on during onboarding.

appcast.xml and HuggingFace model checks are unchanged — they’re per-app, not per-meeting.