StarNet Privacy

Google connector disclosure reviewed: 2026-09-06, against the 0.11.0 candidate. Other sections last reviewed: 2026-07-25.

StarNet is a local-first desktop app. It runs a small server (the "sidecar") on your own machine (localhost) and does the agent work there. Out of the box there is no StarNet account and nothing of yours on a StarNet server — the app talks only to the providers you choose, with your own keys. There is exactly one opt-in exception: if you choose to buy StarNet Credits, that creates an account on our billing service. It is described in full under StarNet Credits below. Optional connections to services such as Google have their own data flows, described below, whether you use your own model key or StarNet Credits. This document is written in plain English and is grounded in an audit of the actual code — not aspirations.

Support questions: androo.agi@gmail.com.

The short version

• The app does not track you. There is no telemetry, no analytics, no crash reporting, or ad SDK. We verified this by searching the whole codebase for the usual suspects (Sentry, PostHog, Mixpanel, Segment, Google Analytics, Amplitude, Datadog): none are present. The desktop app does make the automatic update-manifest request described below; that request carries no user data or app identifier. This stays true whether or not you buy credits — buying credits adds a billing account, not tracking.
• Your saved station data lives on your machine, under your OS user's app-data directory. Content used in agent work can also be sent to your selected model provider and connected services, as described below. Local storage paths are — Windows: %APPDATA%\ai.skynet.harness\workspaces\; macOS: ~/Library/Application Support/ai.skynet.harness/workspaces/; Linux: ~/.local/share/ai.skynet.harness/workspaces/. (The developer-mode sidecar uses %LOCALAPPDATA%\StarNet\ on Windows.) The ai.skynet.harness folder name is an intentional back-compatibility alias kept from before the app was renamed to StarNet on 2026-06-22 — it holds your StarNet data.
• The app talks to the network for the configured or requested work described below, plus the desktop app's automatic update-manifest check. These are the only outbound cases.

What leaves your machine — and only these

StarNet makes outbound network requests in exactly these situations. Nothing else.

1. Your chosen AI model provider (using your key)

When an agent runs, StarNet calls the model provider you configured, authenticated with your API key (or your ChatGPT sign-in). Your prompts, conversation, and any content the agent works with are sent to that provider so it can generate a response — the same as any app that uses that provider. StarNet is a pass-through here; it does not sit in the middle.

Depending on what you set up, that provider is one of:

• OpenRouter (openrouter.ai)
• OpenAI (api.openai.com)
• Anthropic (api.anthropic.com)
• Google Gemini (generativelanguage.googleapis.com)
• ChatGPT via sign-in (chatgpt.com / auth.openai.com)
• or another OpenAI-compatible provider you point it at (xAI, Groq, Mistral, DeepSeek, Together, Fireworks, Perplexity, Cerebras).

Your key, your data, your account. StarNet never sees a StarNet-owned copy — the key is yours and the request goes straight to the provider you picked.

2. Chat channels — only if you connect them

If you connect a chat channel, StarNet talks to that platform to send and receive messages on the channel you set up, authenticated with the token (or endpoint) you provide. Nothing is contacted unless you connect it:

• Discord / Telegram — StarNet calls that platform's API (discord.com, api.telegram.org) with your bot token.
• Slack — StarNet calls the Slack API (slack.com) with your bot token.
• Matrix — StarNet talks to the homeserver you point it at — whatever URL you configure, whether matrix.org or a server you run yourself — using your access token. StarNet does not pick a server; you do.
• Signal — StarNet talks to the signal-cli REST endpoint you run (the URL you configure for your own signal-cli bridge), using the account you registered there.

If you never connect a channel, StarNet never contacts any of these services.

3. Spotify — only if you enable it

If you enable the Spotify integration and authorize it, StarNet calls the Spotify API (api.spotify.com, accounts.spotify.com) to read what's playing and control playback, using the token you granted. If you don't enable Spotify, no Spotify requests are made.

4. Web search / web fetch — only when an agent uses that tool

If an agent uses its web tools, StarNet fetches results through independent, keyless services — web search via Mojeek (mojeek.com, with DuckDuckGo as a fallback) and page reading via Jina Reader (r.jina.ai) — or, if you have an OpenRouter key, OpenRouter's web plugin. If the Jina reader is unavailable or rate-limits, the page is fetched directly from its own host as a fallback, so the site you asked to read may see the request come from your machine. Only your search query or the URL you asked to read is sent, and only when an agent actually runs a web search or fetch. (The fetch tool also refuses to reach private/internal network addresses, as an anti-abuse guard.)

5. Voice — two different paths, and they behave differently

Spoken agent lines use local Kokoro when the built-in voice engine is available. Other spoken lines use your configured model provider (OpenRouter, OpenAI, or Gemini), or ElevenLabs when you configure an ElevenLabs voice, so the line of text is sent to that service. If those paths are unavailable, StarNet can use Microsoft's keyless Edge Read Aloud service, which sends the line of text to Microsoft. Voices off means no TTS requests at all.

Hands-free Live Voice does not use a vendor realtime-voice session. Listening runs on local Whisper when installed; on Windows, offline System.Speech dictation can serve as the fallback. On those paths, your recorded audio never leaves your machine. The open Whisper and Kokoro weights may be fetched from huggingface.co on first use and cached in your user profile; that download carries no audio or transcript. Once cached, listening is offline. Speaking is also offline while Kokoro is available; if it is not, the Edge Read Aloud fallback described above may send only the answer text (not your recorded audio) to Microsoft.

6. A version check for updates (no personal data)

Automatic update checks are on by default. On first run and at the configured interval, the desktop app fetches a single public manifest file from GitHub Releases:

https://github.com/androoAGI/starnet-releases/releases/latest/download/latest.json

This is a plain GET for a static file. No user data, no identifier, and no telemetry are sent — it's the same request your browser would make to view that file. Each update's integrity is cryptographically verified against a key baked into the app before anything is installed. You can stop automatic checks by turning off AUTO-CHECK FOR UPDATES in SYSTEM > SETTINGS > UPDATES; the Update Center still lets you check manually.

7. Google Workspace — only when you connect an account

StarNet's local connector calls the Google services you authorize, using the permissions you approve in Google's consent screen. Gmail, Drive, Calendar, Docs and Sheets are connected separately. Google also supplies the account identity shown with a connected service.

• Gmail: search and read messages, threads and attachments; create drafts and send drafts.
• Drive: search files, read file metadata, export file content, and create or update file metadata within the permissions granted.
• Calendar: read calendars, events and availability.
• Docs and Sheets: read, create and edit documents or spreadsheets.

Use in agent work. Content returned by these tools can become part of the agent's conversation and requests to your chosen model provider. If you use StarNet Credits, those model requests also pass through StarNet's credits gateway. Local execution of the Google connector does not mean that Google-derived content remains exclusively on your device. The provider you choose has its own data-handling terms. Saved conversations, downloaded artifacts and agent memories may retain content from this work.

Credentials and disconnecting. Google sign-in credentials and connector configuration are stored in plaintext JSON under your local StarNet workspace, including recovery copies; these connector credentials are not stored in the OS keychain. Removing a connected service verifies removal of its saved configuration and credentials from the active state and its recovery copy before reporting success. This does not delete prior conversations, artifacts, memories or documents in Google, and it does not itself revoke StarNet's authorization in your Google account. Manage or revoke that authorization separately through your Google account's third-party connections. Revocation stops future authorized access; it does not erase copies already saved locally or sent to a model provider. Delete those separately using the relevant app or provider controls.

What StarNet stores on your machine (and how)

Everything below lives under your per-user app-data directory (see the paths in "The short version" above for Windows/macOS/Linux). It never leaves your machine except as described above. Where this document says "OS keychain," that means Windows Credential Manager, the macOS Keychain, or the Linux Secret Service (e.g. GNOME Keyring), depending on your platform — always under the service name ai.skynet.harness.

Secrets: keychain vs. plaintext — the honest picture

On the desktop build, your provider API keys and your Discord/Telegram bot tokens are held in the OS keychain (Windows Credential Manager, under service ai.skynet.harness), not in a plaintext file. When you upgrade from an older build, any bot token found in the old plaintext channels/secrets.json is migrated into the keychain and stripped from that file.

If you instead run the bare sidecar directly (developer mode / node sidecar/index.js / tests), the OS keychain isn't reachable, so those bot tokens fall back to a plaintext file (channels/secrets.json). This is called out plainly in the code rather than hidden.

These integration secrets are plaintext even on desktop today: the Spotify OAuth token (.secrets/spotify.json) and the ChatGPT/Codex, Grok, and Kimi sign-in tokens (codex/tokens.json, grok/tokens.json, and kimi/tokens.json), plus Google and other connector OAuth credentials (connectors/state.json and its recovery copy). If that matters to you, keep those integrations off. (Transcripts are run through a redaction step at write-time to avoid capturing secret-shaped tokens in your chat history, but the transcript file itself is plaintext.)

Because this data sits in plaintext files under your user profile, anyone with access to your Windows user account can read it. Protect your machine account accordingly.

What we do NOT do

• We do not run analytics or telemetry of any kind.
• We do not collect crash reports.
• We do not require a StarNet account. Unless you buy credits, no account exists, and no StarNet backend holds anything of yours.
• We do not sell your conversations, keys, or files. The app transmits content and authentication information as needed for the model requests and connected-service operations described above.
• We do not store your prompts or your agents' replies on our servers, even on the credits path — see below.

StarNet Credits — only if you buy them

Credits are optional and off by default. If you never buy them, skip this section: nothing in it applies to you, and the app behaves exactly as described above.

If you do buy credits, you create an account on StarNet's billing service, and these things become true:

• What we hold. Your email address, your credit balance and the ledger of grants and charges behind it, an identifier for each station you link, and the customer/subscription identifiers our payment processor gives us. That's the list.
• Card details never reach us. Checkout runs on Stripe, in your browser. StarNet never sees, receives, or stores a card number — the app has no payment form at all.
• How you sign in. By emailed one-time link. There is no StarNet password to steal.
• Model runs go through our gateway. This is the part worth being blunt about: on the credits path your prompts and your agents' replies are relayed through StarNet's servers to the model provider, because that is the only way we can pay for the call on your behalf. We relay them; we do not store their contents, and we do not train on them. What we keep is the metering — model name, token counts, cost, timestamp — which is what your balance is computed from. If you use your own key instead, none of this traffic touches us at all.
• Deleting it. Email androo.agi@gmail.com and we'll delete the account and its data. We keep the minimum transaction records that tax and payment rules require us to keep.

Deleting your data

Your data is just files. To wipe it, uninstall StarNet and delete the ai.skynet.harness app-data folder for your OS — Windows: %APPDATA%\ai.skynet.harness\ (C:\Users\<you>\AppData\Roaming\ai.skynet.harness\); macOS: ~/Library/Application Support/ai.skynet.harness/; Linux: ~/.local/share/ai.skynet.harness/. If you ever ran the developer-mode sidecar on Windows, also delete %LOCALAPPDATA%\StarNet\. Provider API keys and channel tokens held in the OS keychain can be removed there too (search your credential manager / keychain for ai.skynet.harness).

Deleting local files or uninstalling StarNet does not revoke permissions granted in a Google account or remove data already held by Google or a model provider. For connected Google accounts, follow the disconnect and revocation steps above. Copies in backups you made outside the workspace must also be deleted separately.

This website (starnetos.com)

The website you're reading is a static site with no accounts, no cookies, no analytics, and no tracking scripts. Two things are worth stating plainly:

• It's served through Cloudflare, which — like any host/CDN — processes standard request logs (IP address, user agent) to deliver the pages. We add nothing on top.
• The download page makes one anonymous request to api.github.com from your browser to look up the latest release version and exact installer links. GitHub sees that request the same way it would if you visited the releases page yourself.

The "live preview" on the homepage is the real app running entirely in your browser from a seeded demo save; anything you do in it stays in your browser's local storage and goes nowhere.

Changes

If this changes, we'll update this document. Questions: androo.agi@gmail.com.