Session 23 — Website docs read
08-22-2026 → 08-22-2026 · kimi-k3 · 2 messages · 3965 tokens · $0.1334
Summary
Focus: website docs read.
What happened:
- Read RH-WEBSITE.md (website standing plan).
Outcome: operational session only; no changes.
read RH-WEBSITE.md
thinking (226 chars)
The user asked me to read RH-WEBSITE.md. I've read it. Per memory: read ban lifted, but research commits exclude those paths; website edits are the website agent's job. I should just summarize what this file is. Keep it brief.
Read. This is the website ops document — key points:
- **Hard rule (§0):** the website touches only `web/` and `RH-WEBSITE.md`; all research files are read-only (SQLite `mode=ro`, text `'r'`). Commits stage only those two paths.
- **Stack:** Django (`rhweb`/`dashboard`) in `.venv`, served at **127.0.0.1:8001** (8000 reserved), vanilla JS + vendored Plotly/MathJax, 3 s polling of `/api/status/`.
- **13 pages** incl. Dashboard, Roadmap, Ledger, Experiments (charts), Journal, Resources atlas, Live Runs, Feedback, Conversations (agent.db browser).
- **Ops:** session-start = healthcheck `--restart` + watchdog check + review new feedback + one improvement per session (backlog in §7); restart required after any code/template change.
- **Standing constraints** (consistent with my memory): archive renders **verbatim** — no typography/formula transforms (the 2026-08-20 revert); website work is the website agent's job, not the research agent's.
Note the doc is slightly stale vs. the research side — it says "37 D-files" and BUILD STATUS mentions 12 pages, while the archive is now at D-066 and the page list is 13. What would you like to do next?