MCP-Writing-Servers

List all authors

Get author details

Create author

Update author

List books

Get book details

Create book

Update book

Delete a book and all its chapters/scenes

Create chapter

Update chapter

Get chapter details

List chapters

Delete chapter & scenes

Create scene

Update scene

Get scene details

List scenes

Delete scene

Reorder scenes

Analyze scene flow & pacing

Get book structure

Analyze writing progress

Check consistency issues

List/search characters. Provide series_id, search, or both. Unscoped listing not supported.

Get character details

Create a character. Canonical entry point for new characters — a raw database-admin insert on the characters table skips this and won't be visible to knowledge/relationship/continuity tools.

Update core character fields (name, type, status). Prefer this over raw database-admin CRUD so status changes stay consistent with arc and continuity tracking.

Add/update character detail

Get character details

Update character detail

Delete character detail

Record what a character knows, when, and how — maintains knowledge-state continuity. A raw table write bypasses the checks that catch "character knows something before they learned it" plot holes.

Add knowledge w/ chapter ref

Correct an existing character knowledge entry. Identify by `id`. Only provided fields are changed.

Delete a character knowledge entry. Identify by `id`.

Check character knowledge

Find who knows about something

Record a character's presence/state in a chapter or scene — feeds continuity and knowledge-timing checks. A raw insert bypasses those checks.

Get character timeline

Validate a character's state and knowledge stay consistent across a chapter range — catches contradictions a raw CRUD read can't, because it has no notion of "consistent."

Get chapter characters

Query audit logs to review database operations history. Allows filtering by: - Date range (start_date, end_date) - Table name - Operation type (CREATE, READ, UPDATE, DELETE, BATCH_INSERT, BATCH_UPDATE, BATCH_DELETE) - User ID - Success/failure status - Pagination (limit, offset) Use cases: - Compl…

Get comprehensive audit summary statistics. Provides aggregated statistics including: - Total operations count - Success/failure rates - Tables accessed - Unique users - Average and max execution times - Operations breakdown by type - Top tables by activity Useful for: - High-level compliance repo…

Create a full database backup including all tables, schemas, and data. Supports compression and returns backup file details with size, record count, and checksum.

Create a backup of specific database table(s). Supports data-only, schema-only, or full backups with optional compression.

Create an incremental backup containing only changes since the last backup. Requires a base backup and tracks changes via updated_at timestamps.

Export table data to JSON format for data migration or analysis. Supports filtering with WHERE conditions and record limits.

Export table data to CSV format with proper escaping and UTF-8 encoding. Includes column headers and supports filtering.

Restore complete database from a backup file. Validates backup integrity before restore and supports conflict resolution strategies.

Restore a specific table from a table backup file. Validates foreign key constraints and handles record conflicts.

Import data from JSON export file into a table. Validates JSON structure and supports upsert operations.

Import data from CSV file into a table. Supports header row detection, type conversion, and conflict resolution.

List all available backups with metadata including size, date, tables, and record counts. Supports filtering and sorting.

Delete a backup file and its associated manifest. Returns freed disk space. USE WITH CAUTION - deletion is permanent.

Validate backup file integrity by checking checksums, file format, SQL syntax, and dependencies. Returns detailed validation report.

Insert multiple records (1-1000) into a table in a single atomic transaction. All records are inserted or none are (all-or-nothing). Returns all inserted IDs.

Update multiple sets of records in a single atomic transaction. Each update can have different WHERE conditions and data. All updates succeed or all fail (all-or-nothing).

Delete multiple sets of records in a single atomic transaction. Each delete can have different WHERE conditions. All deletes succeed or all fail (all-or-nothing). Supports soft delete.

Query database records with filtering, pagination, and sorting. Supports complex WHERE conditions and multiple sort orders.

Insert a new record into a database table with field validation. Supports all standard PostgreSQL data types.

Update existing records in a database table. Supports partial updates and WHERE conditions to target specific records.

Delete records from a database table. Supports soft-delete (setting deleted_at) or hard delete.

Get detailed schema information for a table including columns, data types, constraints, and indexes. Results are cached for 5 minutes for performance.

List all accessible tables in the database with metadata such as row count estimates and table type. System tables are filtered by default.

Map foreign key relationships for a table, showing which tables it references (parent tables) and which tables reference it (child tables). Supports multi-hop relationship discovery.

Get a lightweight list of column names and types for a table. Useful for autocomplete, suggestions, or quick reference. Faster than db_get_schema.

Gets a board plus its columns (ordered lanes) with per-column card counts — the board-render call.

Lists cards with filters: board, assignee, agent, status, label, priority, agent_claimable_only, include_archived, include_workflow_phase, due_filter, q. The workhorse read tool.

Creates a card — full or quick-add (title-only is valid, the fast path). Board resolves to 'dev-backlog' if neither board_key nor board_id given. review_policy is defaulted by risk class unless explicitly provided.

ATOMIC compare-and-swap claim. Two agents can NEVER both win the same card. NEVER pass an agent id that resolves to an active human identity (see list_identities) — human cards are permanently reserved.

Partial patch of a card — only provided keys change. assignee must be a registered active identity (unknown ids rejected); one resolving to an active human identity auto-clears agent_claimable (DB trigger). assignee:'__clear__' unassigns.

Changes a card's status/lane. Writes activity {action:'moved', from_status, to_status}. A review-required card moving to 'review' is auto-reassigned to 'rebecca'.

Appends to a card comment thread. Agents report progress/results here before moving a card to review.

Lists all boards with their total card counts.

Gets a single card plus its comments, links, activity log, and live workflow_phase if linked — the detail-drawer call.

Attaches a typed link (spec/github_issue/workflow_run/file/url/card) to a card, beyond the inline spec_ref/issue_ref fields.

Archives a card (= move_card to 'archived').

Lists registered kanban identities (human/persona/agent) — the set of ids create_card/update_card will accept as assignee. Defaults to active only. Always excludes the 'test:' id namespace (reserved for throwaway ids minted by tests).

Registers a new identity or updates an existing one's display_name/kind/active. The ONLY sanctioned way to add an assignee value create_card/update_card will accept — there is no silent auto-create on write.

Hard-removes an identity row (e.g. a mis-registered or throwaway id). upsert_identity(active:false) is the softer 'hide from the dropdown' path (list_identities excludes inactive by default) — use delete_identity when the row should not exist at all. Not FK-enforced against kanban_cards.assignee: a …

Get lookup table options

Create lookup option - Admin

Update lookup option - Admin

Delete lookup option - Admin

Assign genres to book

Assign genres to series

Analyze pacing metrics for a chapter including scene distribution, energy modulation, and variance

Analyze pacing across an entire book with aggregated metrics and recommendations

Track stakes escalation in a scene according to NPE Rule #9

Get pressure levels over time for a book to visualize escalation

Log an information reveal according to NPE Rule #8 (only reveal when it alters a choice)

Validate that all information reveals in a book follow NPE Rule #8

Track bidirectional tension between two characters in a scene

Get tension trajectory between two characters across a book

Calculate overall NPE compliance score for a book or chapter

Get all NPE rule violations for a book with severity filtering

Create a new NPE causality chain to track cause-effect relationships in narrative

Add a cause→effect link to an NPE causality chain

Validate an NPE causality chain has no breaks and maintains character agency