MCP Tools Reference

This reference covers all Faker Forge MCP tools for Data Generation and API Mocking.

Tool Groups

• Data generation tools
• Data shaping tools
• Schema management tools
• API mock tools

Data Generation Tools

generate-fake-data

Generate fake data for a new or existing codebase schema.

Required:

• codebase_identifier

Optional:

• schema (full raw schema text)
• additional_context

list-schemas

List available schemas for the authenticated user.

get-codebase-schema

Fetch schema detail by schema_id or codebase_identifier.

Optional:

• include_fields

export-generated-data

Export generated data as a downloadable file and return a signed URL (valid for 1 hour).

Required:

• format: sql, json, csv or xml
• schema_id or codebase_identifier

Optional:

• tables: defaults to every table with generated data
• db_type: mysql (default), mariadb, sqlite or pgsql
• exclude_columns: columns to leave out of this export, for example {"users": ["password"]}

Tables are always written parents first, so SQL imports satisfy foreign keys, including on PostgreSQL.

Data Shaping Tools

These tools give MCP clients the same abilities as the web Data Assistant. Each one:

• requires schema_id, and only works on your own schemas;
• needs a token with the fakerforge:generate ability.

Changes are saved straight away and apply the next time generate-fake-data runs for the schema.

get-schema-overview

Tables with column types and constraints, row counts, which columns have value profiles, relationships (declared foreign keys are marked), and the parent selection. Read-only.

read-schema-source

Read the original SQL of the schema. Read-only.

Optional:

• tables: return only the statements for these tables, including foreign keys added later with ALTER TABLE
• page: page through the whole file, 15,000 characters per page

get-column-profiles

Every column of a table with its faker method, value profile, and whether the profile can be edited. Read-only.

Required:

• table

update-column-profile

Set the value profile of one column. It is validated against the column type, length and CHECK constraints.

Required:

• table, column
• strategy: weighted_set, pool, pattern, numeric, datetime, derived, computed, correlated_group or faker
• params: the strategy's settings, for example {"values": [{"value": "completed", "weight": 80}, {"value": "cancelled", "weight": 20}]}

Keys, foreign keys, ENUM and password columns can't be profiled. See Realistic Values for every strategy.

reprofile-tables

Let the profiling model rewrite whole tables from instructions, for example "Kenyan customers, prices in KES".

Required:

• tables
• instructions

preview-rows

Generate up to 10 sample rows for a table from the current settings without touching stored data. Read-only.

Required:

• table

Optional:

• rows (default 5)

set-row-counts

Required:

• counts: table to row count, for example {"users": 100, "orders": 1000}. A count of 0 removes the table.

set-parent-tables

Choose which parent drives each child table's rows.

Required:

• selections: child to parent, all or none, for example {"orders": "customers"}

manage-relationship

Add, update or remove a relationship used during generation.

Required:

• action: add, update or remove
• child, child_column

Optional:

• parent, parent_column, relationship (one-to-many or one-to-one)
• confirmed: must be true for update and remove, and only after the user has agreed

Foreign keys declared in the schema can't be removed or re-pointed; only their type can change.

Schema Management Tools

add-table-to-schema

Add one or more tables to existing schema context.

remove-table-from-schema

Remove one or more tables from schema context.

add-column-to-schema

Add a column by explicit definition or AI-assisted mode.

remove-column-from-schema

Remove a column from existing schema context.

API Mock Tools

create-api-mock

Create a new mock or add endpoints to an existing mock. The Mock Assistant builds the endpoints from your prompt, saves them straight away and tests them; the response lists the endpoints it created and the assistant's reply. The conversation continues in the mock's Assistant panel.

Required:

• codebase_identifier
• prompt

Optional:

• mock_id
• name
• baseurl

list-mocks

List mock APIs available to the authenticated user.

get-api-mock

Fetch details of a single mock by ID or codebase identifier.

test-mock-endpoint

Simulate which response one of your mocks would return for a request. No HTTP request is sent, nothing is changed, and no delay or quota is used. Returns the matched endpoint and scenario, why it matched, the status, headers and resolved body. Read-only.

Required:

• mock_id
• simulated_method: the HTTP method to match against the mock's endpoints
• path: a concrete path such as /orders/0

Optional:

• query, headers, body: objects, for example {"Authorization": "Bearer x"}

Codebase Identifier Rules

Use this priority order consistently:

1. git remote URL
2. project folder name
3. user-provided project identifier

Consistent identifiers allow schema/mock reuse across sessions.

Common Failure Modes

• Invalid or missing token
• Missing required fields
• Plan limit exceeded
• Polling timeout for long operations
• Unknown schema or mock ID, or a schema owned by another user
• Profile rejected because it doesn't fit the column (type, length, CHECK)
• Relationship change without confirmed: true