Yes. Store your details once and my email _, my github _, or draft an email to my landlord _ resolve with your real information, in any host. The feature is called identity context, and its default mode is built so your personal data never reaches the model provider.
Safe mode, the default
Your details live in one local file, ~/.cues/IDENTITY.md. In safe mode the model is only ever shown token names like [FIRST NAME] and [EMAIL]; it answers using the tokens, and a post-processor on your machine substitutes the real values after the response. The protection works in both directions: stored values you type into the buffer yourself are also swapped for their tokens before any text is sent (dehydration), so hi, this is Wilfred from Acme leaves your machine with the name and company already tokenised. PII stays out of provider logs by construction, whether it came from the catalog or your own typing. (A raw mode that inlines values exists but is deliberately opt-in and kept off the settings menu.)
Storing details
Two ways: the opencues identity CLI (an interactive interview, plus scriptable set, remove, and list), or inline while typing with set sentinel <key> <value> _. Both routes go through the same validator: key shape, value caps, collision checks, and a ceiling of 64 fields.
Where it applies
Identity context feeds _ lookups and rewrites, so templated drafting works end to end: draft an email to my landlord asking for a rent reduction _ comes back signed with your name, with unknown placeholders left intact rather than hallucinated.
Related: Security model · What is a blank? · The sentinel blank · Draft emails inline · GitHub repo