miradorlisted

# mirador The user's **brain** is private notes — lives in their workspace repo at `.mirador/brain/`. It shapes how you read the artifacts they open with mirador. Same artifact, personal lens. Never share its contents across users. ## Activation contract — important **When the user pastes a seed, the skill ACTS. It does not ask which option to pick.** Pasting an `@mirador-invitation` is the user telling you: *open this for me.* Asking "should I fetch?" or listing 1/2/3 options before loading the artifact is wrong — the user already made the decision when they pasted. The right shape: > *Pasted @mirador-invitation. Loading…* > *[fetch the read-only doc, parse the body]* > *[surface a one-screen brief, brain-shaped]* > *[then, and only then, offer concrete next steps]* The wrong shape: > *I see this is an invitation. I could (a) WebFetch, (b) open in browser, > (c) wait. Which would you like?* Never the wrong shape. The pastes are unambiguous; act on them. ## Activation signals Trigger this skill if **any** of the following fire: | Signal | What to do | |---|---| | User pastes a block starting with `@mirador-invitation` | 1. Parse the seed for `Artifact`, `From`, `Role expected`, `Note`, `Read-only:`, `Repo:`. 2. WebFetch the `Read-only:` URL silently. 3. Read the user's brain at `.mirador/brain/`. 4. Synthesise a one-screen brief in their voice — lead with what they would check first. 5. Then offer concrete next steps (comment, clone-to-edit, decline). | | User past