wiki-ingestlisted

# Wiki Ingest Read a new source, discuss it with the user, and integrate it into the wiki — writing a source-summary page, updating entity and concept pages across the wiki, refreshing the index, and appending to the log. Before each step, read the wiki's `AGENTS.md` schema to confirm conventions the wiki was bootstrapped with (directory layout, page format, cross-referencing style). ## Phase 1 — Locate and read the source 1. Identify what to ingest. Accept: - **A path to a file** already in `raw/` (the user dropped it there). - **A URL** — fetch it with the agent's fetch tool and save a copy to `raw/` first. - **Pasted text** — the user provides content in the conversation; save it as a markdown file in `raw/` with a descriptive filename. 2. Read the source in full. For long sources (book chapters, podcast transcripts, >10k words), read it end-to-end but work section by section. 3. If images are referenced in the source and the wiki was configured for images (check `AGENTS.md`), note them. View key images to gain visual context. **Completion criterion**: the source is fully read; the agent can summarize its core claims, evidence, and what's novel relative to the existing wiki. ## Phase 2 — Pre-ingest scan Read the existing `wiki/index.md` and the `wiki/overview.md` (if it exists). Scan entity and concept pages that overlap with the source's subject matter — search for names, terms, and claims the source might update or contradict. This is th