read-up-onlisted

# Read Up On Build a briefing on a topic by discovering and synthesizing everything the vault already knows about it. The user invokes this when they want full context loaded before continuing; they don't know (or care) which files hold the answer. Can be invoked from ANY project on the machine. ## Configuration ``` VAULT="$HOME/dev/personal/vault" ``` All searches below run against `$VAULT`, regardless of the current working directory. The vault's auto-memory lives at `$HOME/.claude/projects/*-dev-personal-vault/memory/MEMORY.md` (glob it; the project dir name encodes the username, which differs between machines). ## When to use Trigger phrases: "read up on ...", "catch me up on ...", "what do I have on ...", "refresh on ...", "load context on ...". Topics can be: - A **project** (e.g. "REMA 1000 tech lead role") - A **person** (e.g. "Øistein", "Daniel Pedersen") - A **concept or domain** (e.g. "agentic commerce", "padel") - A **company/product** (e.g. "Capra", "Shopify") If the request is ambiguous (multiple candidate topics), ask one quick clarifying question before searching. ## Workflow Run steps 1-3 in parallel. Then read, then synthesize. ### 1. Discover candidates (parallel) Cast a wide net. Norwegian + English variants matter; many vault notes are in Norwegian. - **Filename matches**: `find "$VAULT" -iname "*<keyword>*" -type f` for each keyword variant - **Full-text grep**: `grep -r -l -i "<keyword>" --include="*.md" "$VAULT"` for each variant - **Auto