Context intake
What should the Mirror consider?
Draft changes remain in this browser only.
Experimental · local-only reflection surface
What stays true across contexts?
Continuity Mirror is a user-controlled reflection prototype for comparing self-provided context across selected source categories without pretending those categories are connected accounts or authoritative identity records.
Type · Experimental
Implementation · Browser prototype
Storage · Local browser only
Network · No connector calls
Privacy boundaryThis prototype stores only the draft you choose to enter in this browser’s local storage. Source buttons are category selectors, not OAuth connections. Nothing on this page proves an account is linked.
Mirror model
Source ≠ truth.
USER-PROVIDED CONTEXT
EDITABLE SYNTHESIS
NO PROTECTED-TRAIT INFERENCE
EDITABLE SYNTHESIS
NO PROTECTED-TRAIT INFERENCE
Continuity rule
Compare context, surface patterns, preserve uncertainty, and keep the human able to correct the result.
Compare context, surface patterns, preserve uncertainty, and keep the human able to correct the result.
01
Select context
Choose which categories belong in this reflection.
02
Provide context
Use only information you intentionally enter here.
03
Synthesize
Generate editable observations, never hidden certainty.
04
Correct / clear
Revise the draft or remove local state entirely.
Context categories
Select signals to consider.
These controls do not authenticate or connect to any external service. They only label the categories you want represented in the local reflection draft.
Mirror rules
Human correction stays authoritative.
- Do not infer protected or deeply private traits from context categories.
- Use only user-entered or explicitly authorized context.
- Separate public presentation from private identity.
- Describe uncertainty instead of converting patterns into facts.
- Keep every observation editable and removable by the user.
Reflection output
Patterns, not verdicts.
The generated cards are deterministic browser copy built from the local draft. They are not model inference, diagnosis, personality scoring or external account analysis.
Process
Consent → context → synthesis → correction.
A production-grade Mirror would require explicit connector authorization, source-specific scopes, evidence provenance, correction paths and revocation-aware retention. This prototype does not claim those integrations.
Scope
Choose the categories and purpose of the reflection.
Context
Use human-provided text rather than hidden retrieval.
Synthesis
Produce bounded observations with uncertainty intact.
Correction
Revise or delete local state without asking the system for permission.