The privacy model

Privacy isn't a promise here. It's the code.

Second Brain Link runs entirely on your machine. The core makes zero network calls — your exports are never uploaded, and the output is plain Markdown you own. Here's exactly what it does and doesn't do, so you can verify it against the source.

What never leaves your machine

The build pipeline has no network layer. Detecting, resolving, and rendering all happen locally.
No telemetry, no analytics, no "phone home." Run it fully offline.
Where you point your AI afterward is your choice — the vault is just local files.

Scope: "no telemetry, no analytics" describes the downloadable tool (the core pipeline that processes your exports). This marketing website, secondbrainlink.com, uses privacy-respecting, consent-gated Google Analytics — no analytics cookies are set until you accept. See the Privacy Policy.

What gets stripped before anything is written

Third-party (and employee) emails & phone numbers are rejected at parse time — never written.
Message bodies are never read. There is no method that opens a private message; only a frequency/recency signal grades relationship strength.
Sensitive files are quarantined — logins, receipts, and for companies HR / payroll / security — excluded from the build.

The rules aren't a policy page — they're the Collector class. If a rule isn't in the code, it isn't a rule. Read it on GitHub →

sources/common.py

# The privacy policy IS this class.
class Collector:
    def add_person(self, name, **f):
        if EMAIL_RE.match(name) or f.get("phone"):
            return            # PII never written
        self._merge(nk(name), f)

    # there is no add_message_body(). by design.
    def add_message_signal(self, name, date):
        self.sig[nk(name)].bump(date)

QUARANTINE = {"Logins", "Receipts", "messages", "HR", "Payroll"}

What you own at the end

Plain Markdown and a standard Obsidian vault. No proprietary format, no lock-in.
Delete it, fork it, grep it. It's just files in a folder.
Re-run any time from a fresh export. Need everything for your own brain? Owner mode (--full) keeps it all — still local.

Not affiliated with or endorsed by LinkedIn, Meta, Google, Slack, Obsidian, Anthropic, OpenAI, or the GBrain project. Works with the data exports those services provide to you, and emits to the open-source GBrain format.

Try Studio

GitHub & Skills

For

Privacy isn't a promise here. It's the code.

What never leaves your machine

What gets stripped before anything is written

What you own at the end