A 5e storytelling engine with an LLM DM

Complete plan 0007 and queue streaming markdown, then DM prompting master

The knowledge mount held up in a live DeepInfra session: lookup, read, and a wikilink hop from lore into a module statblock, all streaming. Plan 0007 moves to completed. The next two phases come straight from that session: the DM's markdown renders as raw asterisks, so a streaming markdown renderer is phase 6, and the DM told the player what Mother Uldra is, so phase 7 takes on prompting and the DM-versus-player knowledge line. Context stack and persistence shift to phases 8 and 9. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01CUXjWo1zGFhdJUig1hcQGf


+14 -5
2 changed files