Analyze TDS student discussions for the **most recently completed Wednesday–Tuesday week (IST)**.

Use @LocalMCP2 to read:

- **Discourse:** Refresh `~/code/exam/dumps/discourse-YYYY.md` using `~/code/exam/dumps/update_discourse.py` with credentials from `~/code/exam/.env`.
- **WhatsApp:** `~/Documents/data/whatsapp/TDS _ Sept 2026 [[email protected]].jsonl`
- **Google Chat:** Space `spaces/AAQAO77CWr8`, via `gws` with `GOOGLE_WORKSPACE_CLI_CONFIG_DIR=~/.config/[email protected]`. Use the Chat API, not Gmail.

Use relevant @Anand Skills, especially **reframe-question, data-analysis, blind-spot, evidence-provenance, and verification-gate**.

Help me understand:

1. **What students discuss:** Major themes, concerns, excitement, surprises, and how these evolve.
2. **How students learn together:** Concrete stories of collaboration, teaching, troubleshooting, building, sharing, gaming the system, and helping strangers.
3. **What educators and experimenters can learn:** Intended vs. actual learning, unintended incentives, usability problems, novel behaviors, and experiments worth trying.
4. **Interesting tidbits:** Funny, touching, surprising, creative, or revealing side conversations that might otherwise go unnoticed.

Prioritize **BIG, USEFUL, SURPRISING** insights over exhaustive topic summaries. Look for counterexamples, quiet successes, emerging patterns, and what is *not* being discussed. Distinguish genuine learning from merely earning marks.

Ground findings in specific conversations, with dates, brief quotes, and source links where available. Separate reported problems from verified defects. Avoid letting a few prolific participants dominate the narrative. Anonymize students where appropriate.

Start with a compelling executive summary, followed by evidence-rich stories, notable tidbits, and 3–5 actionable lessons or experiments. Include a compact source coverage summary, highlighting stale or missing data.

Write an engaging, concise weekly report (~800–1,200 words), suitable for sharing with the TDS teaching team. Don't invent evidence or manufacture insights. Don't modify course systems or publish anything.