Ghostlight/notes/ghostlight-implementation-plan.md
2026-05-02 12:19:33 +01:00

73 lines
4.5 KiB
Markdown

# Ghostlight Implementation Plan
## Current Phase
Build the first reliable data-generation loop for socially persistent Aetheria
agents. The immediate target is not a full simulation. It is a clean, reviewed,
sandboxed training-data pipeline for branching scenes and state consequences.
## Near-Term Sequence
1. Stabilize canonical state and projection seams.
- Keep `schemas/agent-state.schema.json` and required vectors coherent.
- Keep canonical state separate from perceived overlays.
- Keep voice, presentation, relationship, memory, and situational pressure explainable.
- Do not leak raw numeric state internals into responder prompt text.
2. Stabilize responder packet and output seams.
- Use `schemas/responder-packet.schema.json` and `schemas/responder-output.schema.json`.
- Build packets from coordinator artifacts plus projected local context.
- Preserve exact responder-visible input, hidden-context audit, allowed actions, source excerpts, output contract, and lore-access mode.
- Preserve raw output, parsed output, review labels, consulted refs, research summary, leakage audit, and coordinator interventions.
3. Tighten research-enabled responder capture.
- The first no-fork research-enabled Sella sample and reviewed mutation receipt now exist.
- Preserve `consulted_refs`, `followed_refs`, `research_trace`, and `research_summary` inside raw responder output when the responder or runner produces them; mirror indexable provenance into capture metadata.
- Treat research scope as seed docs plus bounded relevant traversal, not a four-file pinhole.
- Require PSC, territory, resident-faction, technology, and social-movement context when it materially affects the scene.
- Require `runner_captured` research trace status for accepted research-enabled gold data; coordinator-reconstructed trace is useful draft audit, not proof of the responder's actual research path.
- The Veyr/Callisto provenance sample is the first accepted coordinator-scoped retrieval lane with `runner_captured` trace. Use it as the current pattern for trace-backed gold candidates.
- Continue rejecting archived local-model prototype runners for gold responder data unless the task is explicitly model plumbing.
4. Generate the next reviewed sample.
- Start from `examples/agent-state.cold-wake-story-lab.after-veyr-category-offer.json`.
- Build the next projected local context and responder packet from reviewed state, not from coordinator memory.
- Run the next no-fork sandboxed responder/coordinator step with exact visible input, raw output, runner-captured research trace, leakage audit, and review labels.
- Convert accepted output into mutation receipts.
- Keep fuzzy social changes manual and auditable until appraiser/classifier models exist.
5. Generalize the loop.
- Add more historical grounded fixtures from AetheriaLore.
- Add future-branch Elysium fixtures with branch lineage and constraint labels.
- Track coverage across every major faction, minor faction, movement, and
major flashpoint; major factions require both founding-era and
day-in-the-life stories.
- Require cultural collision in coverage stories so training data captures
inter-faction dynamics, mutual misreads, and movement pressure instead of
sealed faction portraits.
- Emit technology/item manifest deltas when scenes discover or stress gear, assemblies, supply chains, or faction tech bases.
- Keep artifacts database-shaped enough for future game-engine integration.
6. Train only after schemas stop sliding.
- Pilot samples are schema shakedown, not robust corpus scale.
- Specialized future models include coordinator, retriever, projector,
responder, appraiser, mutator, relationship/perception updater, evaluator,
and institution/faction/consumer decision models.
- Deterministic gates remain code: visibility, action legality, object
custody, resource accounting, schema validation, source provenance,
mutation authority, prompt leakage checks, and Ink compilation.
## Deferred
- Full world simulation loop
- Autonomous offscreen faction simulation
- Economy simulation loop
- Long-horizon plot generation without author scaffolding
- Fine-tuning before artifact schemas, review criteria, and evaluators stabilize
## Discipline
- Work one bounded organ at a time.
- Prefer explicit maps and contracts over implicit context.
- If the diff grows while understanding shrinks, stop and simplify.
- Keep history in git and archived receipts; keep persisted state focused on the mission.