Publishing Checklist

<!-- translations:start -->

한국어 · 中文 · 日本語 · Русский · Español · Français · Deutsch

<!-- translations:end --> Use this checklist before presenting Tesserae publicly.

Repository hygiene

• [ ] README explains what the project is and what problem it solves.
• [ ] Install command works from a fresh shell.
• [ ] Quickstart uses tesserae, not python3 -m.
• [ ] Architecture docs explain raw evidence → graph → projections.
• [ ] Feature map lists implemented features without overselling future work.
• [ ] Session-history docs explain explicit import, privacy review, generated routes, and transcript typography.
• [ ] Self-dogfood demo has been run and documented.
• [ ] Generated artifacts are reproducible and either ignored or intentionally published.
• [ ] RAG-Anything index refreshed (if enabled)

Verification

PYTEST_DISABLE_PLUGIN_AUTOLOAD=1 python3 -m pytest tests/ -q
./scripts/install.sh --help
tesserae project setup --help
tesserae project compile --help

Self-dogfood

tesserae project setup \
  --yes \
  --name tesserae_self \
  --source README.md \
  --source docs \
  --source tesserae \
  --source tests \
  --source scripts \
  --with-understand-anything \
  --install-understand-anything \
  --understand-anything-platform codex \
  --with-raganything \
  --install-raganything \
  --raganything-parser mineru \
  --run-raganything \
  --run-cognee \
  --install-cognee
tesserae project compile
tesserae project sessions list
tesserae project build-site
tesserae project serve --port 8765

Demo talking points

• Tesserae is not a generic noun-phrase graph. It uses a controlled ontology.
• Research and development code share infrastructure but keep distinct schemas.
• Markdown and HTML are projections, not authoritative truth stores.
• The default path is local and no-API-key friendly.
• Agent harnesses and MCP make the graph usable by coding agents.
• Imported harness session pages turn prior Claude Code/Codex work into searchable project memory while keeping transcript discovery explicit.