Read cohorts; channel_pid required
| Name | Required | Description | Default |
|---|---|---|---|
| date_to | Yes | ||
| date_from | Yes | ||
| event_keys | No | ||
| channel_pid | Yes | ||
| product_ref | No | ||
| product_name | No |
| Name | Required | Description | Default |
|---|---|---|---|
| message | No | ||
| retryable | No | ||
| support_ref | No |
Changes observed during successful MCP inspections.
Output schema / $defs / saAdded value: +{
+ "properties": {
+ "can_report_stored_total": {
+ "type": "boolean"
+ },
+ "reason": {
+ "type": "string"
+ },
+ "zero_proven": {
+ "type": "boolean"
+ }
+ },
+ "required": [
+ "can_report_stored_total",
+ "zero_proven",
+ "reason"
+ ],
+ "type": "object"
+}Output schema / else / properties / read_evidenceAdded value: +{
+ "description": "Proofs apply independently to returned metrics and scope as stored/observed. A new read_ref is a call identifier, never a source watermark. False or missing proof means unknown, not zero. Fingerprints correlate requests/results in audits; they are not authorization or pagination tokens.",
+ "properties": {
+ "artifact_returned": {
+ "type": "boolean"
+ },
+ "configuration": {
+ "properties": {
+ "account_inventory_enumerated": {
+ "const": false
+ },
+ "all_campaigns_paused_proven": {
+ "const": false
+ },
+ "requested_ids_complete": {
+ "type": "boolean"
+ },
+ "scope": {
+ "const": "requested_entity_ids_only"
+ }
+ },
+ "type": "object"
+ },
+ "mmp": {
+ "properties": {
+ "can_report_scoped_totals": {
+ "type": "boolean"
+ },
+ "provider_cache_max_age_seconds": {
+ "const": 90
+ },
+ "reason": {
+ "type": "string"
+ },
+ "scope": {
+ "const": "linked_apps_queried_channels_and_windows"
+ },
+ "source_observed_at": {
+ "type": "null"
+ }
+ },
+ "type": "object"
+ },
+ "read_ref": {
+ "pattern": "^read_[0-9a-f]{32}$",
+ "type": "string"
+ },
+ "request_fingerprint": {
+ "type": "string"
+ },
+ "result_fingerprint": {
+ "type": "string"
+ },
+ "scopes": {
+ "items": {
+ "properties": {
+ "all_campaigns_paused_proven": {
+ "const": false
+ },
+ "funnel": {
+ "additionalProperties": {
+ "$ref": "#/$defs/sa"
+ },
+ "maxProperties": 32,
+ "type": "object"
+ },
+ "index": {
+ "type": "integer"
+ },
+ "inventory_anchor": {
+ "type": [
+ "string",
+ "null"
+ ]
+ },
+ "inventory_coverage": {
+ "type": "string"
+ },
+ "inventory_source_observed_at": {
+ "type": [
+ "string",
+ "null"
+ ]
+ },
+ "metrics": {
+ "additionalProperties": {
+ "$ref": "#/$defs/sa"
+ },
+ "type": "object"
+ },
+ "metrics_source_observed_at": {
+ "type": [
+ "string",
+ "null"
+ ]
+ },
+ "range_complete": {
+ "type": [
+ "boolean",
+ "null"
+ ]
+ },
+ "scope": {
+ "const": "stored_query"
+ }
+ },
+ "type": "object"
+ },
+ "maxItems": 10,
+ "type": "array"
+ }
+ },
+ "required": [
+ "read_ref",
+ "request_fingerprint",
+ "result_fingerprint"
+ ],
+ "type": "object"
+}Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
Annotations already declare readOnly, non-destructive, non-open-world, so the safety profile is covered. The description adds the mandatory channel_pid requirement, which is useful context beyond the annotations, but says nothing about rate limits, data freshness, or what the query returns.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
Extremely short, which is efficient, but the terseness veers into under-specification rather than disciplined conciseness. One sentence plus a fragment leaves the agent without enough to act confidently.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
With 6 undocumented parameters, a required-channel constraint, and no explanation of what a "product cohort" query returns, the description is incomplete for a query tool at this complexity level. An output schema exists, so return shape needn't be described, but the input semantics gap remains.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Schema description coverage is 0% across 6 parameters, so the description must carry the semantic burden. It only flags one required param (channel_pid) and says nothing about date_from/date_to, event_keys, product_ref, or product_name.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
"Read cohorts" gives a verb and a resource, but "cohorts" is vague and the description never mentions MMP, products, or events, which are the distinguishing features of this tool. It does not differentiate from siblings like mmp_insights_get_product_today or mmp_insights_query_product_event_summary.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
No when-to-use guidance, no alternatives named, no mention of what kind of cohort analysis this enables. The only concrete signal is a required-parameter note.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Add one secure layer between your agents and this server.