README.md

Documentation Index

• docs/fragments-and-prose-chain.md
  • Fragment system and prose chain: complete schema reference, fragment types, ID conventions, prose chain structure, filesystem storage layout, full API reference, import/export format, SillyTavern character card import (PNG + JSON with lorebook/world book support), and porting guide.
• docs/context-blocks.md
  • Context block system: structured LLM prompt blocks, the beforeBlocks plugin hook, Block Editor UI, custom blocks (simple + script), and block configuration API.
• docs/summarization-and-memory.md
  • Technical reference for source-linked story memory: contribution currency, target-relative projection, explicit coverage gaps, reader-specific presentation, authored memory, prompt budgeting, and performance properties.
• docs/summary-projection-design.md
  • Current tiered-memory contract: source-current contributions, recursive content-addressed roll-ups, target-safe frontier selection, retrospective register, coverage spans, explicit gaps, and reader-specific presentation.
• docs/analyze-context-design.md
  • Analyze context strategy: character working-set tiering, explicit tool lanes, and edit-safety via proposals against preloaded sheets.
• docs/context-provenance-and-continuity.md
  • Architecture for context receipts, source-linked continuity projections, chronology, open-thread lifecycle, character knowledge, and authority boundaries.
• docs/timelines.md
  • Timelines (branches): isolated story copies, forking, switching, storage layout, and API.
• docs/character-chat.md
  • Character Chat architecture and API: conversation storage model, NDJSON streaming chat endpoint, persona modes, model/provider resolution, and frontend integration points.
• docs/prose-writing-panel.md
  • Prose Writing Panel flow: Tiptap-based editor behavior, autosave rules, selection transform operations (rewrite/expand/compress/custom), and the librarian/prose-transform API.
• docs/backend-stress-harness.md
  • How to run the synthetic backend stress harness for summary/index scalability and context build latency benchmarking.
• docs/third-party-plugins.md
  • How to build external plugins, use runtime iframe UI, scaffold from templates, client-side panel hooks (onPanelOpen/onPanelClose for bundled, postMessage for iframe), query cache invalidation, and validate in dev.
• docs/runtime-plugins-and-binary-packaging.md
  • Runtime plugin loading, binary build/package workflow, CI release automation, and deployment notes.
• docs/component-ids.md
  • Frontend data-component-id contract for extensibility and automation.
• docs/publishing-plugin-sdk.md
  • How to publish @viscerous/errata-plugin-sdk manually and via GitHub Actions.
• docs/instruction-registry.md
  • Instruction registry: centralized built-in prompt defaults, registered keys, and the per-agent block customization path.
• docs/generation-pipeline.md
  • Generation pipeline: standard and prewriter modes, context assembly, streaming events, generation logs, and mode-specific behavior.
• docs/adding-agents.md
  • Contributor guide for adding new agents: file structure, system prompts, block builders, runtime logic, registration, model resolution, UI integration, testing checklist.
• docs/agent-observability.md
  • Agent observability: active registry, live activity stream, run traces, and the persistent librarian analyses/generation logs — which system answers which question, and who wires them (you don't).
• docs/documentation-sync-skill.md
  • Implementation notes and usage for the commit-driven documentation sync skill.

Related:

• plugins/templates/README.md
  • Plugin recipe template catalog.
• packages/errata-plugin-sdk/package.json
  • Local SDK package: @viscerous/errata-plugin-sdk.

Automation:

• bun run docs:sync
  • Rebuilds docs/commit-doc-sync.md from git history and stores state in .agent/docs-sync-state.json.