fix(mcp): hide hog function inputs and reject no-field exclusions

docs/published/handbook/engineering/ai/implementing-mcp-tools.md

@@ -327,7 +327,7 @@ Product teams own their definitions and control which operations are exposed as
        list: true # marks as a list endpoint
        enrich_url: '{id}' # appended to url_prefix for result URLs
        category: Other product # overrides the file-level category, e.g. for $mcp_tool_category in MCP analytics
-       exclude_params: [field] # hide params from tool input
+       exclude_params: [field, steps.*.secret] # hide params and body fields from tool input (`*` steps into array items and dict values)
        include_params: [field] # whitelist params (excludes all others)
        response: # filter response fields (applied per-item on list endpoints)
          include: [id, key, name] # keep only these fields (dot-path wildcards supported)
@@ -357,6 +357,7 @@ Product teams own their definitions and control which operations are exposed as
    The MCP tool then requires the field, even when the generated PATCH body marks it optional.
 
    Unknown keys are rejected at build time (Zod `.strict()`) to catch typos early.
+   An `exclude_params` entry that names no query parameter or request body field also fails the build, so a typo cannot leave the field exposed.
 
    For list tools with a UI app, set `response.text_include` to the dot-path fields an agent
    needs from each result row. This adds a compact text response for clients that cannot read

products/autoresearch/mcp/tools.yaml

@@ -44,14 +44,6 @@ tools:
             target_event, or from the action name when target_event is omitted, if you do not set it. It must be unique
             among the project's non-archived pipelines. After creating, call autoresearch-train-create to start an
             asynchronous agent training run.
-        exclude_params:
-            - id
-            - created_by
-            - created_at
-            - updated_at
-            - last_scored_at
-            - status
-            - iteration_budget_remaining
         feature_flag: autoresearch
     autoresearch-list:
         operation: autoresearch_list
@@ -208,12 +200,6 @@ tools:
             autoresearch-create with name set to suggested_name and the same five fields plus output_person_property.
             For an action target, pass the same target_definition to both calls; target_event then only labels it, and
             both calls need the action:read scope.
-        exclude_params:
-            - resolved_activity_event
-            - activity_event_alternatives
-            - display_name
-            - description
-            - notes
         feature_flag: autoresearch
     autoresearch-resume-create:
         operation: autoresearch_resume_create
@@ -319,12 +305,6 @@ tools:
             to see how the suggestion was interpreted. Examples: 'try a gradient boosting model', 'remove recency
             features — I suspect leakage', 'the target event has a 7-day spike on Mondays, account for day-of-week'.
             Suggestions survive pipeline pauses and are included in the next training batch when the pipeline resumes.
-        exclude_params:
-            - source
-            - status
-            - agent_response
-            - created_by
-            - linked_iteration_ids
         feature_flag: autoresearch
     autoresearch-suggestions-list:
         operation: autoresearch_suggestions_list

products/batch_exports/mcp/tools.yaml

@@ -27,16 +27,7 @@ tools:
             convertToProjectTimezone, for example {"convertToProjectTimezone": true} to export them in the project
             timezone.
         exclude_params:
-            - id
-            - team_id
-            - created_at
-            - last_updated_at
-            - last_paused_at
-            - latest_runs
             - filters
-            - start_at
-            - end_at
-            - integration
     batch-export-delete:
         operation: batch_exports_destroy
         enabled: true
@@ -83,14 +74,8 @@ tools:
             and Redshift. The destination type cannot be changed. Do not change the model field on an existing batch
             export.
         exclude_params:
-            - id
-            - team_id
-            - created_at
-            - last_updated_at
-            - last_paused_at
-            - latest_runs
             - filters
-            - integration
+
     batch-exports-list:
         operation: batch_exports_list
         enabled: true

products/billing/mcp/tools.yaml

@@ -73,9 +73,6 @@ tools:
             period. For time-series breakdowns of usage call billing-usage-get; for cost (spend) breakdowns call
             billing-spend-get. To check whether the customer has access to a specific feature (e.g. group analytics,
             white-labelling), inspect `available_product_features` in the response.
-        exclude_params:
-            - limit
-            - offset
         url_prefix: /organization/billing/overview
         response:
             exclude:

products/dashboards/mcp/tools.yaml

@@ -466,18 +466,6 @@ tools:
             remove it. To add new widget tiles to this dashboard, use dashboard-widgets-batch-add instead. The returned
             tiles omit insight results to save context — use dashboard-insights-run to fetch the actual data for each
             insight.
-        exclude_params:
-            - creation_mode
-            - effective_restriction_level
-            - effective_privilege_level
-            - is_shared
-            - access_control_version
-            - last_refresh
-            - last_accessed_at
-            - deleted
-            - persisted_filters
-            - persisted_variables
-            - _create_in_folder
         param_overrides:
             id:
                 cast: string-int

products/persons/mcp/tools.yaml

@@ -144,4 +144,3 @@ tools:
             'plan', 'role', or 'company'. Provide the property key and optionally a search value to filter results.
         exclude_params:
             - format
-            - force_refresh

products/warehouse_sources/mcp/tools.yaml

@@ -62,9 +62,6 @@ tools:
             schemas (Stripe, Zendesk) but on a large production database (a Postgres/MySQL/BigQuery/Snowflake with many
             tables) it syncs them all, which can be costly and noisy. When the user wants only specific tables, or
             non-default sync types per table, use 'external-data-sources-create' instead.
-        exclude_params:
-            - client_secret
-            - account_id
         param_overrides:
             source_type:
                 input_schema: ExternalDataSourceTypeSchema
@@ -346,9 +343,6 @@ tools:
             warehouse tables for revenue (Stripe), CRM (Hubspot, Salesforce), support (Zendesk), and your own databases
             (Postgres, MySQL, BigQuery).
         exclude_params:
-            - client_secret
-            - account_id
-            - job_inputs
             - created_via
         inject_body:
             created_via: mcp

services/mcp/scripts/generate-tools.ts

@@ -76,6 +76,7 @@ interface OpenApiSchema {
     readOnly?: boolean
     writeOnly?: boolean
     items?: OpenApiSchema | { $ref: string }
+    additionalProperties?: OpenApiSchema | { $ref: string } | boolean
     properties?: Record<string, OpenApiSchema>
     required?: string[]
     $ref?: string
@@ -502,6 +503,55 @@ interface SchemaComposition {
     paramAliases: Record<string, string[]>
 }
 
+function assertExclusionsNameRealFields(
+    name: string,
+    config: ToolConfig,
+    resolved: ResolvedOperation,
+    spec: OpenApiSpec
+): void {
+    const queryParamNames = new Set(
+        (resolved.operation.parameters ?? []).filter((p) => p.in === 'query').map((p) => p.name)
+    )
+    const bodySchema = resolved.operation.requestBody?.content?.['application/json']?.schema
+    for (const entry of config.exclude_params ?? []) {
+        if (!queryParamNames.has(entry) && !schemaHasPath(spec, bodySchema, entry.split('.'))) {
+            throw new Error(
+                `Enabled tool "${name}": exclude_params entry "${entry}" names no query parameter or body field, ` +
+                    'so it hides nothing. Fix the name or remove the entry.'
+            )
+        }
+    }
+}
+
+function schemaHasPath(
+    spec: OpenApiSpec,
+    schemaOrRef: OpenApiSchema | { $ref: string } | undefined,
+    segments: string[]
+): boolean {
+    const schema = schemaOrRef && resolveSchema(spec, schemaOrRef)
+    const [head, ...tail] = segments
+    if (!schema || head === undefined) {
+        return false
+    }
+    const variants = [...(schema.oneOf ?? []), ...(schema.anyOf ?? []), ...(schema.allOf ?? [])]
+    if (variants.some((variant) => schemaHasPath(spec, variant, segments))) {
+        return true
+    }
+    if (head === '*') {
+        return (
+            tail.length > 0 &&
+            [schema.items, schema.additionalProperties].some(
+                (child) => typeof child === 'object' && schemaHasPath(spec, child, tail)
+            )
+        )
+    }
+    const property =
+        schema.properties && Object.prototype.hasOwnProperty.call(schema.properties, head)
+            ? schema.properties[head]
+            : undefined
+    return tail.length === 0 ? property !== undefined : schemaHasPath(spec, property, tail)
+}
+
 function composeToolSchema(
     config: ToolConfig,
     resolved: ResolvedOperation,
@@ -1833,6 +1883,7 @@ function generateCategoryFile(
             process.exit(1)
         }
         try {
+            assertExclusionsNameRealFields(name, config, resolved, spec)
             enabledTools.push([
                 name,
                 {

services/mcp/tests/unit/__snapshots__/tool-schemas/cdp-functions-create.json

@@ -104,22 +104,9 @@
         "inputs": {
             "additionalProperties": {
                 "properties": {
-                    "bytecode": {
-                        "items": {},
-                        "type": "array"
-                    },
                     "bytecode_contract": {
                         "type": "string"
                     },
-                    "order": {
-                        "type": "number"
-                    },
-                    "templating": {
-                        "description": "* `hog` - hog\n* `liquid` - liquid",
-                        "enum": ["hog", "liquid"],
-                        "type": "string"
-                    },
-                    "transpiled": {},
                     "value": {}
                 },
                 "type": "object"
@@ -285,22 +272,9 @@
                             "inputs": {
                                 "additionalProperties": {
                                     "properties": {
-                                        "bytecode": {
-                                            "items": {},
-                                            "type": "array"
-                                        },
                                         "bytecode_contract": {
                                             "type": "string"
                                         },
-                                        "order": {
-                                            "type": "number"
-                                        },
-                                        "templating": {
-                                            "description": "* `hog` - hog\n* `liquid` - liquid",
-                                            "enum": ["hog", "liquid"],
-                                            "type": "string"
-                                        },
-                                        "transpiled": {},
                                         "value": {}
                                     },
                                     "type": "object"