Consent-gated · built and tested

Your Privacy.
Your Memory.
Your Control.

Ownership is a mechanism here, not a promise. Consent is checked before recall, writes are arithmetic so text cannot become a belief, and the person a profile describes can read it, correct it, or erase it — from the population prior as well as the graph.

Consent-gated everywhere

Recall is checked against consent before the engine will answer at all.

Glass-box, not black box

Every stored preference is visible and editable, with why() showing the evidence behind it.

Injection-resistant by construction

Writes are arithmetic, so page or user text cannot be talked into becoming a belief.

Apache-2.0, auditable

The engine, the tests, and the benchmarks are public. Read them before trusting them.

Consent GateChecked before recallYour StoreSQLite or PostgresOpen SourceApache-2.0Right to ForgetPrior includedGlass-Box EditorSee and fix everything

How We Protect You

Consent-Gated Recall

Recall is blocked unless consent covers the purpose being asked for.

Your Store, Your Server

SQLite by default or Postgres in production. Whoever deploys it holds the data.

Site Isolation

Cross-site assembly is opt-in and default-deny, with sensitive categories walled off.

Cascading Erasure

forget_everywhere clears the profile and unlearns the person from the population prior.

Tamper-Evident Log

Actions are recorded in an HMAC chain the user can replay to detect alteration.

Private Collective Priors

Cold-start borrows crowd patterns under k-anonymity and bounded-mean differential privacy.

Your Data, Your Rules

These are engine operations, not settings screens: each maps to a REST endpoint and the glass-box editor shipped in the repository.

Read the code

Review

Open the glass-box editor and read every stored preference.

Export

Take the whole profile out through the REST export endpoint.

Correct

Edit a preference the engine got wrong, and it stays corrected.

Forget

Erase a single memory or the entire profile, prior included.

Privacy is not a feature. It’s our foundation.

FERNme is an Apache-2.0 research preview at v0.4, with 283 passing tests.
Published results are synthetic or LLM-authored; a real-human pilot is the pending next step.