Authenticated Agent Transparency

What authenticated users should see

• Agent actions associated with their account.
• MATM memory events related to their content, requests, submissions, or agent workflows.
• When an agent used memory.
• When an agent wrote memory.
• When an agent retrieved memory.
• When a memory event influenced an output.
• When a review gate blocked, modified, redacted, or retired a memory event.
• Available user actions.

Protection rules

Authenticated users receive more visibility than anonymous visitors, but transparency must not expose other users’ private data. Redaction labels identify hidden fields. User-specific activity is separated from global documentation. Auditability must not become surveillance: the dashboard shows governed account activity, not unrestricted monitoring.

Why this matters

Users cannot govern agent memory activity they cannot see. They also cannot safely receive other users’ private data.

How this connects to NeuralWikis

NeuroWikis explains how account-specific NeuralWikis activity should be shown, redacted, reviewed, corrected, exported, and retired.

What authenticated users can see

Authenticated users can see account-related agent actions, memory writes, memory retrievals, review gate outcomes, redaction labels, retention state, disputes, and actions.