anecdoteslisted

# /anecdotes — Collect news evidence for a project When invoked, gather news/anecdotal evidence relevant to the current project's research topic and store it under `references/news/` using the canonical newsbr layout. ## What you do 1. **Read project context** to figure out what to search for: - `<project>/CLAUDE.md` (current focus, key terms) - `<project>/docs/summary.md` (research question) - `<project>/paper/main.tex` introduction, if present - `<project>/README.md` 2. **Decide if the project is Brazil-focused.** Most projects in this workspace are. If non-Brazil, expand outlet selection accordingly (see "Non-Brazil mode" below). 3. **Propose 3–8 search queries** covering different angles of the topic. Show them to the user before running. Briefly explain why each query. Wait for confirmation unless the user said "go ahead" up front. 4. **Run newsbr collect** for each approved query, against the relevant outlets: ```bash python3 -m newsbr collect \ --project <project_root> \ --query "<query>" \ --outlets migalhas,folha,estadao,jota \ --max-pages 1 \ --context "/anecdotes for <project>" ``` Always pass `--context` so the audit log captures who/why. 5. **Report back** what was added: number of new articles per outlet, total added, and the path to `search_log.md`. If you saw notable articles, mention 2–3 by title. 6. **Update `docs/notes/anecdotes.md`** (create if missing) with a brief one-line