gbrain-ingestlisted

# Ingest Skill ## Workflow 1. **Read the source.** Meeting transcript, article, document, conversation log. Identify: participants, companies, topics, decisions, action items. 2. **For each entity mentioned:** - `gbrain search "<name>"` — does a page exist? - **If yes:** `gbrain get <slug>` to read current compiled_truth. Rewrite State section with new info. Append to timeline. `cat updated.md | gbrain put <slug>` with updated content. - **If no:** Create page using the appropriate template (see Page Templates below). `cat new.md | gbrain put <slug>` with new content. 3. **Extract and create links.** - For every entity-to-entity reference in the source: `gbrain link <from> <to> --context "the sentence containing the reference"` - Links are stored directionally. Create both directions if both pages exist and the relationship is mutual. 4. **Parse timeline entries.** - For each datable event: `gbrain timeline-add <slug> --date YYYY-MM-DD --summary "one line" --source "meeting/doc-ref"` 5. **Tag appropriately.** - `gbrain tag <slug> <tag>` — use lowercase, hyphenated tags (e.g. yc-alum, investor, active-deal) 6. **Verify.** - `gbrain stats` — confirm page counts look right. - `gbrain get <slug>` — spot-check a few pages. ## Entry criteria Not everything gets a page. The bar: - Anyone you met 1:1 or in a small group: YES - People discussed in substantive context: YES - Companies in deal or investment context: YES - Concep