Ghostlight

Ghostlight is where the stranger agent work gets to live.

The project is about socially persistent generative agents: layered personality, cultural priors, masks, memory, misreadings, incentives, and prompt projection that grows out of structured latent state instead of one wet persona paragraph trying to hold up the ceiling.

The immediate goal is not a finished runtime. The immediate goal is a repo that can carry the architecture cleanly enough that the runtime does not become a beautiful little superstition.

What Lives Here

• Epiphany-style persistent state surfaces for re-entry, compaction, and distilled evidence.
• Architecture notes for agent state, signal classification, culture-shaped distributions, and prompt projection.
• Aetheria authoring targets for Call of the Void content generation, dialogue scaffolding, and procedural drama across narrated timeline stories.
• A bounded working discipline that keeps maps, scratch, evidence, and plans from dissolving into one confident soup.

Repo Tour

• AGENTS.md: operating discipline for humans and agents
• state/map.yaml: canonical project map
• state/scratch.md: disposable local working memory
• state/evidence.jsonl: distilled belief-changing evidence
• notes/fresh-workspace-handoff.md: compact re-entry packet
• notes/ghostlight-current-system-map.md: source-grounded explanation of the live repo machinery
• notes/ghostlight-implementation-plan.md: current larger-organ sequence
• notes/architecture-rationale.md: why the state split exists at all
• docs/research/: research backlog and modeling notes
• docs/aetheria/: imported or local Aetheria source material used by Ghostlight
• docs/architecture/: world, state, and prompt-system architecture notes
• tools/ghostlight_state.py: compact state helper
• tools/ghostlight_prepare_compaction.py: pre-compaction audit helper

Quick Commands

npm run state:status
npm run state:prepare-compaction
& 'C:\Users\Meta\.cache\codex-runtimes\codex-primary-runtime\dependencies\python\python.exe' '.\tools\ghostlight_state.py' add-evidence --type research --status ok --note '...'

If you are steering the repo through Codex, read AGENTS.md first. That is the operating manual. The README is just the sign on the door.

Related Docs

• docs/research/things-to-steal.md
• docs/research/personality-model-roadmap.md
• docs/research/personality-signal-classifier.md
• docs/aetheria/call-of-the-void-brainstorm.md
• docs/architecture/agent-state-distributions-and-prompt-projection.md
• docs/architecture/aetheria-persistent-agent-architecture.md
• docs/architecture/aetheria-content-generation-targets.md
• docs/architecture/aetheria-agent-population-model.md