Skip to content

FEATURES.md

Latest commit

 

History

History
180 lines (114 loc) · 8.33 KB

FEATURES.md

File metadata and controls

180 lines (114 loc) · 8.33 KB

Writing MCP — Features

Shipped product capabilities and the initiative docs that explain them in more detail.

Use this page when you want to understand what the product can do today. Use PRODUCT.md for the high-level product overview, active direction, backlog, and foundations.

Core Platform

🎯 Metadata Architecture & Ownership

How metadata is stored, managed, and kept in sync with prose.

Highlights:

  • Tier 1 (structural) and Tier 2 (editorial) metadata split
  • SQLite-canonical structural and relationship metadata with .meta.yaml compatibility input/output
  • Auto-migration from legacy frontmatter
  • Staleness detection and enrich_scene

📦 Import & Sync Operations

How manuscripts are imported from Scrivener and synced into the indexed workspace.

Highlights:

  • SQLite index with universe/project/scene/character/place/thread schema
  • Scrivener binder-ID based identity
  • World folder structure for characters, places, and reference docs
  • Stale metadata warnings on sync

✏️ Prose Editing & Version Control

Two-step editing workflow with git-backed history.

Highlights:

  • propose_edit to commit_edit confirmation flow
  • Pre-edit snapshots before every commit
  • Manual snapshots for restore points
  • Git-backed version history rather than database snapshots

🔍 Search, Querying & Analysis

Fast metadata-first discovery with prose loaded on demand.

Highlights:

  • find_scenes() with metadata filters
  • get_arc() for ordered scene journeys
  • search_metadata() with FTS5
  • Staleness warnings before analysis

🧾 Review Bundles for Editorial Workflows

Deterministic bundle generation for outline discussion, detailed editing, and beta reading.

Highlights:

  • preview_review_bundle planning step
  • create_review_bundle artifact generation
  • PDF export with manifest and review companion files
  • Strictness modes for stale or incomplete metadata

🔐 Beta Reader Accountability and Book-Like Layout

Chapter-scoped beta packets with per-page accountability and improved PDF reading ergonomics.

Highlights:

  • chapters filter support for one/few chapter beta bundles
  • Per-page PDF footer accountability with recipient and fingerprint token
  • Manifest fingerprint metadata for provenance and traceability
  • 6x9 book-like PDF geometry for beta profile readability

🗂️ Scrivener Direct Extraction

Direct ingestion from .scriv and .scrivx internals for richer metadata extraction.

Highlights:

  • Official direct binder ingestion path
  • Richer metadata than External Folder Sync alone
  • Safeguards to avoid schema-coupled regressions

📚 Reference Document Querying

Reference note indexing and linkage for world-building, continuity, and research material.

Highlights:

  • Query reference docs alongside manuscript metadata
  • Link reference docs to scenes
  • Support continuity and rules-of-the-world questions

Recently Delivered

🧭 Architecture Alignment Follow-up

Target-architecture follow-up work for managed sync, sidecar compatibility, and outcome-oriented relationship workflows.

Highlights:

  • Ordinary managed-project sync preserves canonical structure and relationship rows when filesystem representations are missing
  • Generic metadata updates preserve structural sidecar compatibility fields unless an explicit structure or repair workflow owns the change
  • Relationship work is routed through outcome-level tools for arcs, character/place evidence, relationship beats, reference evidence, and metadata audits
  • Sidecar compatibility and migration docs explain retained import, generated, review, and recovery roles without making sidecars daily-work authority
  • Release-log entries for PRs #222, #224, #225, and #226 document the delivered migration posture and user-facing guidance

💾 Database Backup and Recovery

Git-reviewable backup artifacts and explicit restore workflows for SQLite-canonical project state.

Highlights:

  • export_project_backup writes deterministic manifest.json, canonical.snapshot.json, and advisory operations.jsonl artifacts under project-backups/<project_id>/
  • diagnose_project_backups reports missing, stale, wrong-project, incompatible, tampered, partial, and unreadable backup bundles
  • Sanctioned canonical mutation tools refresh backup artifacts and operation history after successful project-scoped changes
  • restore_project_from_backup defaults to dry-run planning, then applies trusted restore plans transactionally with explicit destructive and cross-scope confirmations
  • Backup and Recovery explains backup layers, Git expectations, Docker/homeserver volume backups, and restore failure handling

🧱 Structural Authority Hardening

Final authority hardening after the target architecture migration.

Highlights:

  • Ordinary sync observes sidecar and folder drift without adopting it as canonical structure for managed projects
  • update_scene_metadata rejects structural fields and routes chapter placement/order changes to explicit structure workflows
  • Structure export diagnostics report missing, stale, wrong-project, and incompatible-schema exports before repair
  • restore_structure_from_export provides explicit transactional recovery from trusted generated exports
  • Numeric chapter inputs remain read-scope compatibility aliases resolved through canonical chapter identity

🏗️ Target Architecture Migration

Structural manuscript state boundaries, explicit mutation workflows, and SQLite-canonical structure storage.

Highlights:

  • Structure inference, sidecar structural writes, sync observation, diagnostics, and compatibility resolution now have clearer internal boundaries
  • Explicit structure commands cover scene assignment, chapter creation/rename/reorder, epigraph attachment, and scene movement
  • SQLite is documented as canonical for structural manuscript state while prose remains file-based
  • export_structure_snapshot generates deterministic SQLite-derived structure exports for Git review and future recovery workflows

📚 Chapter and Epigraph Structure

First-class chapters and epigraphs with canonical chapter_id targeting, explicit structure workflows, chapter-aware retrieval, and chapter-linked rendering.

Highlights:

  • Canonical chapters and epigraphs entities with project-scoped identifiers
  • list_chapters and find_epigraphs as dedicated discovery tools
  • find_scenes, get_chapter_prose, styleguide flows, and review bundles updated for chapter_id
  • Explicit chapter-folder and epigraph.md indexing path, with numeric chapter filters retained as read-scope compatibility aliases resolved through canonical chapter identity
  • Explicit chapter/scene/epigraph structure workflows and managed-sync guardrails keep SQLite canonical structure authoritative

Implementation status:

Follow-up work:

🪄 Guideline Generation

Reusable prose styleguide system with config resolution, skill generation, and edit-time enforcement behavior.

Follow-up work:

Additional Completed References