Manufact

item-brain

mcp.item.app/mcp

The official Item MCP server. Ask a read-only Item assistant, search and mutate internal CRM data, manage Item lists/views/notes/tasks, and inspect object schemas. External messaging and integration side-effect tools are not exposed. Connect with Item OAuth. Legacy ?api_key auth is available only when explicitly enabled for migration.

Protocol: MCP 2025-11-25
Server version: 1.0.0
Supports: Tools
Resources
Prompts
Logging
Completions

Overall score

Last reviewed at Oct 14, 2026, 16:00:07 UTC

Server metadata

4 of 8 fields50/1005 / 10 pts

Identity, discovery, and usage metadata advertised during the MCP handshake.

Stable server name

item-brain

10/10Source

Human-readable title

Add a human-readable title for clients and directories.

0/10Source

Useful description

20/20Source

Server version

Version 1.0.0

10/10Source

Public website

Link to a public website or documentation page.

0/10Source

Scalable or high-resolution icon

Advertise a vector or at least 128px server icon.

0/10Source

Server instructions

Add concise instructions that explain how an agent should use the server.

0/20Source

Declared capabilities

Supports Tools, Resources, Prompts, Logging, Completions.

10/10Source

Protocol

Pass100/10020 / 20 pts

Connection, authentication, and protocol compatibility.

MCP negotiation succeeds (v1/v2)

MCP legacy negotiation succeeded in 2150ms

Source

Tools list returns results

22 tool(s) returned

Source

Resources list responds

0 resource(s) returned

Source

Server declares tools capability

Server declares tools capability in initialize response

Source

Tools

Warnings95/10028.5 / 30 pts

Tool definitions, annotations, and client compatibility.

All tools have descriptions

All tools have non-empty descriptions

Skills over MCP resources are readable

Server does not advertise the Skills over MCP extension

Claude Code metadata fits its 2 KB fields

1 metadata field(s) exceed Claude Code's 2 KB truncation point

How to fix: Keep each tool description and the server instructions field at or below 2 KB, and put critical guidance first.

warningclaude-code

Source

All tools have OpenAI behavior hints

All tools declare the three OpenAI behavior hints as booleans

Source

Tool names use valid characters

All tool names use valid characters

Source

Claude tools declare applicable behavior hints

All tools include an applicable readOnlyHint or destructiveHint

Source

All tools have a display title

Source

Claude tool names are within 64 characters

All tool names are within Claude's 64-character limit

Source

Cursor tool-name budget advisory

No tool name independently exhausts Cursor's observed 60-character combined budget

Observed behavior

Tool names follow the MCP 1–128 character range

All tool names follow the MCP recommended length

Source

Tool names follow MCP uniqueness guidance

All tool names are unique

Source

ChatGPT tool names are unique

All tool names are unique

Source

Tool input header annotations are valid

All x-mcp-header annotations satisfy Streamable HTTP requirements

Source

Tool input schemas have a valid object structure

All tool input schemas have a valid object structure

Source

OpenAI file parameters use valid schemas

All declared OpenAI file parameters use reviewer-valid schemas

Source

Tool text has no hidden control characters

No invisible or control characters found in tool-facing text

Source

Declared output schemas have a valid basic structure

0 declared output schema(s) pass basic structural validation; schemas may describe any JSON value, including arrays

Source

MCP Apps

Not applicable-Optional

App resources, metadata, and rendering requirements.

All discovered view resources were inspected

No view resources found — view checks skipped

Source

Tool-linked resource URI is readable

No view resources found — view checks skipped

Source

Tool invocation metadata present

No view resources found — view checks skipped

Source

View resource envelope is valid

No view resources found — view checks skipped

View resource metadata complete

No view resources found — view checks skipped

Source

Tool visibility metadata is valid

No view resources found — view checks skipped

Source

Declared domains respond

No view resources found — view checks skipped

CSP metadata covers literal external origins

No view resources found — view checks skipped

Source

View resources declare a domain

No view resources found — view checks skipped

Source

Tool documentation quality

22 tools reviewed74/10029.6 / 40 pts

How clearly each tool’s title, description, input and output schemas, and behavior annotations guide an agent.

manage_listTitles 96Descriptions 84Inputs 0Annotations 75Outputs 056/100

Title96/100

Manage an Item list

The title clearly communicates the tool’s action and subject. Keep it concise.

Description84/100

Create, read, rename or edit the description of a static Item list; add or remove up to 1,000 explicitly selected member IDs; or delete the list definition without deleting its records. IDs are strings. Deletion requires explicit user confirmation and confirmed=true. Existing object type and sharing are preserved; dynamic lists and object-type changes are not supported. Membership calls are atomic: an invalid member fails the whole call. Use Item import jobs for per-record outcomes and resumable large imports.

The description explains the capability, when to use it, and what to expect.

Inputs0/100

0 of 23 inputs have descriptions. Add descriptions to the missing inputs, including where to obtain IDs and how to supply values.

operation › oneOf › 0 › action

Not provided

Missing description. Explain what this value means and how to supply it.

operation › oneOf › 0 › object

Not provided

Missing description. Explain what this value means and how to supply it.

operation › oneOf › 0 › name

Not provided

Missing description. Explain what this value means and how to supply it.

operation › oneOf › 0 › description

Not provided

Missing description. Explain what this value means and how to supply it.

operation › oneOf › 1 › action

Not provided

Missing description. Explain what this value means and how to supply it.

operation › oneOf › 1 › object

Not provided

Missing description. Explain what this value means and how to supply it.

operation › oneOf › 1 › after

Not provided

Missing description. Explain what this value means and how to supply it.

operation › oneOf › 1 › limit

Not provided

Missing description. Explain what this value means and how to supply it.

operation › oneOf › 2 › action

Not provided

Missing description. Explain what this value means and how to supply it.

operation › oneOf › 2 › listId

Not provided

Missing description. Explain what this value means and how to supply it.

operation › oneOf › 2 › after

Not provided

Missing description. Explain what this value means and how to supply it.

operation › oneOf › 2 › limit

Not provided

Missing description. Explain what this value means and how to supply it.

operation › oneOf › 3 › action

Not provided

Missing description. Explain what this value means and how to supply it.

operation › oneOf › 3 › listId

Not provided

Missing description. Explain what this value means and how to supply it.

operation › oneOf › 3 › name

Not provided

Missing description. Explain what this value means and how to supply it.

operation › oneOf › 3 › description

Not provided

Missing description. Explain what this value means and how to supply it.

operation › oneOf › 4 › action

Not provided

Missing description. Explain what this value means and how to supply it.

operation › oneOf › 4 › listId

Not provided

Missing description. Explain what this value means and how to supply it.

operation › oneOf › 4 › confirmed

Not provided

Missing description. Explain what this value means and how to supply it.

operation › oneOf › 5 › action

Not provided

Missing description. Explain what this value means and how to supply it.

operation › oneOf › 5 › listId

Not provided

Missing description. Explain what this value means and how to supply it.

operation › oneOf › 5 › object

Not provided

Missing description. Explain what this value means and how to supply it.

operation › oneOf › 5 › objectIds

Not provided

Missing description. Explain what this value means and how to supply it.

Annotations75/100

Declare idempotentHint based on the tool’s actual behavior.

{
  "title": "Manage an Item list",
  "readOnlyHint": false,
  "openWorldHint": false,
  "destructiveHint": true
}

Output schema0/100

No output schema is declared. Add one for structured results so clients can interpret returned fields.

Not declared

Original tool metadata

{
  "name": "manage_list",
  "title": "Manage an Item list",
  "description": "Create, read, rename or edit the description of a static Item list; add or remove up to 1,000 explicitly selected member IDs; or delete the list definition without deleting its records. IDs are strings. Deletion requires explicit user confirmation and confirmed=true. Existing object type and sharing are preserved; dynamic lists and object-type changes are not supported. Membership calls are atomic: an invalid member fails the whole call. Use Item import jobs for per-record outcomes and resumable large imports.",
  "inputSchema": {
    "type": "object",
    "$schema": "http://json-schema.org/draft-07/schema#",
    "required": [
      "operation"
    ],
    "properties": {
      "operation": {
        "oneOf": [
          {
            "type": "object",
            "required": [
              "action",
              "object",
              "name"
            ],
            "properties": {
              "name": {
                "type": "string",
                "maxLength": 255,
                "minLength": 1
              },
              "action": {
                "type": "string",
                "const": "create"
              },
              "object": {
                "type": "string",
                "minLength": 1
              },
              "description": {
                "anyOf": [
                  {
                    "type": "string",
                    "maxLength": 2000
                  },
                  {
                    "type": "null"
                  }
                ]
              }
            },
            "additionalProperties": false
          },
          {
            "type": "object",
            "required": [
              "action"
            ],
            "properties": {
              "after": {
                "type": "string",
                "default": "0",
                "pattern": "^[0-9]+$"
              },
              "limit": {
                "type": "integer",
                "default": 100,
                "maximum": 100,
                "minimum": 1
              },
              "action": {
                "type": "string",
                "const": "list"
              },
              "object": {
                "type": "string"
              }
            },
            "additionalProperties": false
          },
          {
            "type": "object",
            "required": [
              "action",
              "listId"
            ],
            "properties": {
              "after": {
                "type": "string",
                "default": "0",
                "pattern": "^[0-9]+$"
              },
              "limit": {
                "type": "integer",
                "default": 100,
                "maximum": 100,
                "minimum": 1
              },
              "action": {
                "type": "string",
                "const": "get"
              },
              "listId": {
                "type": "string",
                "pattern": "^[1-9][0-9]*$"
              }
            },
            "additionalProperties": false
          },
          {
            "type": "object",
            "required": [
              "action",
              "listId"
            ],
            "properties": {
              "name": {
                "type": "string",
                "maxLength": 255,
                "minLength": 1
              },
              "action": {
                "type": "string",
                "const": "update"
              },
              "listId": {
                "type": "string",
                "pattern": "^[1-9][0-9]*$"
              },
              "description": {
                "anyOf": [
                  {
                    "type": "string",
                    "maxLength": 2000
                  },
                  {
                    "type": "null"
                  }
                ]
              }
            },
            "additionalProperties": false
          },
          {
            "type": "object",
            "required": [
              "action",
              "listId",
              "confirmed"
            ],
            "properties": {
              "action": {
                "type": "string",
                "const": "delete"
              },
              "listId": {
                "type": "string",
                "pattern": "^[1-9][0-9]*$"
              },
              "confirmed": {
                "type": "boolean",
                "const": true
              }
            },
            "additionalProperties": false
          },
          {
            "type": "object",
            "required": [
              "action",
              "listId",
              "object",
              "objectIds"
            ],
            "properties": {
              "action": {
                "enum": [
                  "add",
                  "remove"
                ],
                "type": "string"
              },
              "listId": {
                "type": "string",
                "pattern": "^[1-9][0-9]*$"
              },
              "object": {
                "type": "string",
                "minLength": 1
              },
              "objectIds": {
                "type": "array",
                "items": {
                  "type": "string",
                  "pattern": "^[1-9][0-9]*$"
                },
                "maxItems": 1000,
                "minItems": 1
              }
            },
            "additionalProperties": false
          }
        ]
      }
    }
  },
  "annotations": {
    "title": "Manage an Item list",
    "readOnlyHint": false,
    "openWorldHint": false,
    "destructiveHint": true
  }
}

ask_item_assistantTitles 93Descriptions 86Inputs 88Annotations 75Outputs 079/100

Title93/100

Ask Item assistant

The title clearly communicates the tool’s action and subject. Keep it concise.

Description86/100

Ask the Item assistant a natural-language question in read-only mode. Prefer this tool for complex requests that require analysis, multi-step reasoning, synthesis across object types, emails, meetings, meeting recordings, timelines, or a large number of CRM objects. Use the direct structured tools only for simple deterministic lookups or explicit internal Item mutations. Thread continuity is important: if a previous ask_item_assistant response returned thread_id, always pass the most recent thread_id on every follow-up so the Item assistant can continue the same conversation. Omit thread_id only when intentionally starting a brand-new Item assistant thread.

The description explains the capability, when to use it, and what to expect.

Inputs88/100

Remove validation rules already expressed in this field’s schema; describe its meaning and usage instead.

message

User message for the Item assistant.

Remove validation rules already expressed in this field’s schema; describe its meaning and usage instead.

thread_id

Existing Item assistant thread ID to continue. If any previous ask_item_assistant call in this MCP conversation returned a thread_id, always pass the most recent thread_id on follow-up calls. Omit only when intentionally starting a brand-new Item assistant thread.

Clear enough to supply this argument.

Annotations75/100

Declare idempotentHint based on the tool’s actual behavior.

{
  "title": "Ask Item assistant",
  "readOnlyHint": true,
  "openWorldHint": true,
  "destructiveHint": false
}

Output schema0/100

No output schema is declared. Add one for structured results so clients can interpret returned fields.

Not declared

Original tool metadata

{
  "name": "ask_item_assistant",
  "title": "Ask Item assistant",
  "description": "Ask the Item assistant a natural-language question in read-only mode. Prefer this tool for complex requests that require analysis, multi-step reasoning, synthesis across object types, emails, meetings, meeting recordings, timelines, or a large number of CRM objects. Use the direct structured tools only for simple deterministic lookups or explicit internal Item mutations. Thread continuity is important: if a previous ask_item_assistant response returned thread_id, always pass the most recent thread_id on every follow-up so the Item assistant can continue the same conversation. Omit thread_id only when intentionally starting a brand-new Item assistant thread.",
  "inputSchema": {
    "type": "object",
    "$schema": "http://json-schema.org/draft-07/schema#",
    "required": [
      "message"
    ],
    "properties": {
      "message": {
        "type": "string",
        "minLength": 1,
        "description": "User message for the Item assistant."
      },
      "thread_id": {
        "type": "string",
        "format": "uuid",
        "pattern": "^([0-9a-fA-F]{8}-[0-9a-fA-F]{4}-[1-8][0-9a-fA-F]{3}-[89abAB][0-9a-fA-F]{3}-[0-9a-fA-F]{12}|00000000-0000-0000-0000-000000000000|ffffffff-ffff-ffff-ffff-ffffffffffff)$",
        "description": "Existing Item assistant thread ID to continue. If any previous ask_item_assistant call in this MCP conversation returned a thread_id, always pass the most recent thread_id on follow-up calls. Omit only when intentionally starting a brand-new Item assistant thread."
      }
    }
  },
  "annotations": {
    "title": "Ask Item assistant",
    "readOnlyHint": true,
    "openWorldHint": true,
    "destructiveHint": false
  }
}

execute_viewTitles 98Descriptions 78Inputs 83Annotations 75Outputs 077/100

Title98/100

Run saved view

The title clearly communicates the tool’s action and subject. Keep it concise.

Description78/100

Execute an accessible saved Item view with its configured filters, sorting, and columns. Results are paginated.

Describe the returned information so the agent knows whether it will answer the request.

Inputs83/100

Remove validation rules already expressed in this field’s schema; describe its meaning and usage instead.

view_id

The UUID of the view to execute

Clear enough to supply this argument.

limit

Maximum number of results (default: 50, max: 200)

Remove validation rules already expressed in this field’s schema; describe its meaning and usage instead.

offset

Number of results to skip for pagination (default: 0)

Remove validation rules already expressed in this field’s schema; describe its meaning and usage instead.

Annotations75/100

Declare idempotentHint based on the tool’s actual behavior.

{
  "title": "Run saved view",
  "readOnlyHint": true,
  "openWorldHint": false,
  "destructiveHint": false
}

Output schema0/100

No output schema is declared. Add one for structured results so clients can interpret returned fields.

Not declared

Original tool metadata

{
  "name": "execute_view",
  "title": "Run saved view",
  "description": "Execute an accessible saved Item view with its configured filters, sorting, and columns. Results are paginated.",
  "inputSchema": {
    "type": "object",
    "$schema": "http://json-schema.org/draft-07/schema#",
    "required": [
      "view_id"
    ],
    "properties": {
      "limit": {
        "type": "number",
        "description": "Maximum number of results (default: 50, max: 200)"
      },
      "offset": {
        "type": "number",
        "description": "Number of results to skip for pagination (default: 0)"
      },
      "view_id": {
        "type": "string",
        "description": "The UUID of the view to execute"
      }
    }
  },
  "annotations": {
    "title": "Run saved view",
    "readOnlyHint": true,
    "openWorldHint": false,
    "destructiveHint": false
  }
}

search_objectsTitles 96Descriptions 73Inputs 74Annotations 75Outputs 072/100

Title96/100

Search CRM objects

The title clearly communicates the tool’s action and subject. Keep it concise.

Description73/100

Structured read-only search over Item CRM objects. Use for deterministic object lookup by query, object type, field, relationship, limit, and offset.

Resolve contradictions between the description, input schema, and annotations.

Inputs74/100

Explain the meaning of this value, not just its name.

query

Search query. Use "*" to return all records.

Clear enough to supply this argument.

object_slug

Object type slug, such as contacts, companies, deals.

Clear enough to supply this argument.

field_name

Optional field to search in.

Explain the meaning of this value, not just its name.

parent_object_slug

Parent object slug for relationship traversal.

Clear enough to supply this argument.

parent_object_id

Parent object ID for relationship traversal.

Clarify units, where to obtain this value, or conventions not already expressed in the schema.

relationship_field_name

Relationship field on the parent object type.

Explain the meaning of this value, not just its name.

limit

Maximum number of records to return (1-100).

Remove validation rules already expressed in this field’s schema; describe its meaning and usage instead.

offset

Number of records to skip, for pagination.

Remove validation rules already expressed in this field’s schema; describe its meaning and usage instead.

Annotations75/100

Declare idempotentHint based on the tool’s actual behavior.

{
  "title": "Search CRM objects",
  "readOnlyHint": true,
  "openWorldHint": false,
  "destructiveHint": false
}

Output schema0/100

No output schema is declared. Add one for structured results so clients can interpret returned fields.

Not declared

Original tool metadata

{
  "name": "search_objects",
  "title": "Search CRM objects",
  "description": "Structured read-only search over Item CRM objects. Use for deterministic object lookup by query, object type, field, relationship, limit, and offset.",
  "inputSchema": {
    "type": "object",
    "$schema": "http://json-schema.org/draft-07/schema#",
    "properties": {
      "limit": {
        "type": "integer",
        "maximum": 100,
        "minimum": 1,
        "description": "Maximum number of records to return (1-100)."
      },
      "query": {
        "type": "string",
        "description": "Search query. Use \"*\" to return all records."
      },
      "offset": {
        "type": "integer",
        "maximum": 9007199254740991,
        "minimum": 0,
        "description": "Number of records to skip, for pagination."
      },
      "field_name": {
        "type": "string",
        "description": "Optional field to search in."
      },
      "object_slug": {
        "type": "string",
        "description": "Object type slug, such as contacts, companies, deals."
      },
      "parent_object_id": {
        "type": "number",
        "description": "Parent object ID for relationship traversal."
      },
      "parent_object_slug": {
        "type": "string",
        "description": "Parent object slug for relationship traversal."
      },
      "relationship_field_name": {
        "type": "string",
        "description": "Relationship field on the parent object type."
      }
    }
  },
  "annotations": {
    "title": "Search CRM objects",
    "readOnlyHint": true,
    "openWorldHint": false,
    "destructiveHint": false
  }
}

get_objectTitles 96Descriptions 92Inputs 50Annotations 75Outputs 071/100

Title96/100

Get CRM object

The title clearly communicates the tool’s action and subject. Keep it concise.

Description92/100

Fetch one item CRM object by object_slug and object_id. Returns the same object shape as search_objects. Contacts and companies also include system_fields: readable System Field metadata (timestamps, counts, generalized labels) such as last email, last/next meeting, and last phone call. Answer value-only questions from it; it never contains content.

The description explains the capability, when to use it, and what to expect.

Inputs50/100

Explain the meaning of this value, not just its name.

object_slug

Object type slug.

Explain the meaning of this value, not just its name.

object_id

Object ID.

Explain the meaning of this value, not just its name.

Annotations75/100

Declare idempotentHint based on the tool’s actual behavior.

{
  "title": "Get CRM object",
  "readOnlyHint": true,
  "openWorldHint": false,
  "destructiveHint": false
}

Output schema0/100

No output schema is declared. Add one for structured results so clients can interpret returned fields.

Not declared

Original tool metadata

{
  "name": "get_object",
  "title": "Get CRM object",
  "description": "Fetch one item CRM object by object_slug and object_id. Returns the same object shape as search_objects. Contacts and companies also include system_fields: readable System Field metadata (timestamps, counts, generalized labels) such as last email, last/next meeting, and last phone call. Answer value-only questions from it; it never contains content.",
  "inputSchema": {
    "type": "object",
    "$schema": "http://json-schema.org/draft-07/schema#",
    "required": [
      "object_slug",
      "object_id"
    ],
    "properties": {
      "object_id": {
        "type": "integer",
        "maximum": 9007199254740991,
        "minimum": -9007199254740991,
        "description": "Object ID."
      },
      "object_slug": {
        "type": "string",
        "description": "Object type slug."
      }
    }
  },
  "annotations": {
    "title": "Get CRM object",
    "readOnlyHint": true,
    "openWorldHint": false,
    "destructiveHint": false
  }
}

list_object_typesTitles 100Descriptions 77Inputs 100Annotations 75Outputs 081/100

Title100/100

List object types

The title clearly communicates the tool’s action and subject. Keep it concise.

Description77/100

List active system and custom object types available to this item organization. Call this tool before searching for objects to get the list of available object types.

Describe the returned information so the agent knows whether it will answer the request.

Inputs100/100

This tool declares no inputs, so argument descriptions are not needed.

No input fields.

Annotations75/100

Declare idempotentHint based on the tool’s actual behavior.

{
  "title": "List object types",
  "readOnlyHint": true,
  "openWorldHint": false,
  "destructiveHint": false
}

Output schema0/100

No output schema is declared. Add one for structured results so clients can interpret returned fields.

Not declared

Original tool metadata

{
  "name": "list_object_types",
  "title": "List object types",
  "description": "List active system and custom object types available to this item organization. Call this tool before searching for objects to get the list of available object types.",
  "inputSchema": {
    "type": "object",
    "$schema": "http://json-schema.org/draft-07/schema#",
    "properties": {}
  },
  "annotations": {
    "title": "List object types",
    "readOnlyHint": true,
    "openWorldHint": false,
    "destructiveHint": false
  }
}

get_object_schemaTitles 98Descriptions 78Inputs 50Annotations 75Outputs 068/100

Title98/100

Get object schema

The title clearly communicates the tool’s action and subject. Keep it concise.

Description78/100

Get field definitions for one item object type by slug. Use before constructing structured filters, field-aware UI, or performing field updates.

Describe the returned information so the agent knows whether it will answer the request.

Inputs50/100

Explain the meaning of this value, not just its name.

object_slug

Object type slug.

Explain the meaning of this value, not just its name.

Annotations75/100

Declare idempotentHint based on the tool’s actual behavior.

{
  "title": "Get object schema",
  "readOnlyHint": true,
  "openWorldHint": false,
  "destructiveHint": false
}

Output schema0/100

No output schema is declared. Add one for structured results so clients can interpret returned fields.

Not declared

Original tool metadata

{
  "name": "get_object_schema",
  "title": "Get object schema",
  "description": "Get field definitions for one item object type by slug. Use before constructing structured filters, field-aware UI, or performing field updates.",
  "inputSchema": {
    "type": "object",
    "$schema": "http://json-schema.org/draft-07/schema#",
    "required": [
      "object_slug"
    ],
    "properties": {
      "object_slug": {
        "type": "string",
        "description": "Object type slug."
      }
    }
  },
  "annotations": {
    "title": "Get object schema",
    "readOnlyHint": true,
    "openWorldHint": false,
    "destructiveHint": false
  }
}

create_objectTitles 95Descriptions 80Inputs 79Annotations 75Outputs 075/100

Title95/100

Create CRM object

The title clearly communicates the tool’s action and subject. Keep it concise.

Description80/100

Create a new object in the system. objectSlug is required, and include all field values for the object. Duplicate guard for contacts and companies: before calling this tool, search internally for an existing record. If you used enrichment, search internally AGAIN using enriched identifiers such as domain, website, email, company name, or LinkedIn URL, and only create if that second search still finds no clear match. If the task requested a restricted optional field, omit its value and list only its field name in restrictedFieldNames. A restricted required field blocks creation.

Explain mutations, irreversible effects, or prerequisites that matter before calling this tool.

Inputs79/100

Remove validation rules already expressed in this field’s schema; describe its meaning and usage instead.

objectSlug

Object type slug, such as contacts, companies, or deals.

Clear enough to supply this argument.

name

Display name for the new object.

Remove validation rules already expressed in this field’s schema; describe its meaning and usage instead.

fields › items › name

Field name.

Remove validation rules already expressed in this field’s schema; describe its meaning and usage instead.

fields › items › value

Field value.

Explain the meaning of this value, not just its name.

restrictedFieldNames

Names of restricted fields requested by the task that were intentionally omitted. Never include their values.

Clear enough to supply this argument.

profileImageUrl

Image URL - e.g. LinkedIn profile picture or the company logo. Do not pass anything here if the object is not a contact or company (it will be ignored).

Clear enough to supply this argument.

_skip_approval

Task runs only: set true to auto-execute without approval when this action is routine and low-risk.

Clear enough to supply this argument.

Annotations75/100

Declare idempotentHint based on the tool’s actual behavior.

{
  "title": "Create CRM object",
  "readOnlyHint": false,
  "openWorldHint": false,
  "destructiveHint": false
}

Output schema0/100

No output schema is declared. Add one for structured results so clients can interpret returned fields.

Not declared

Original tool metadata

{
  "name": "create_object",
  "title": "Create CRM object",
  "description": "Create a new object in the system. objectSlug is required, and include all field values for the object.\n\nDuplicate guard for contacts and companies: before calling this tool, search internally for an existing record. If you used enrichment, search internally AGAIN using enriched identifiers such as domain, website, email, company name, or LinkedIn URL, and only create if that second search still finds no clear match.\n\nIf the task requested a restricted optional field, omit its value and list only its field name in restrictedFieldNames. A restricted required field blocks creation.",
  "inputSchema": {
    "type": "object",
    "$schema": "http://json-schema.org/draft-07/schema#",
    "required": [
      "objectSlug",
      "name",
      "fields"
    ],
    "properties": {
      "name": {
        "type": "string",
        "minLength": 1,
        "description": "Display name for the new object."
      },
      "fields": {
        "type": "array",
        "items": {
          "type": "object",
          "required": [
            "name"
          ],
          "properties": {
            "name": {
              "type": "string",
              "minLength": 1,
              "description": "Field name."
            },
            "value": {
              "description": "Field value."
            }
          },
          "additionalProperties": false
        },
        "description": "Field values to set on the new object, as name/value pairs."
      },
      "objectSlug": {
        "type": "string",
        "minLength": 1,
        "description": "Object type slug, such as contacts, companies, or deals."
      },
      "_skip_approval": {
        "type": "boolean",
        "description": "Task runs only: set true to auto-execute without approval when this action is routine and low-risk."
      },
      "profileImageUrl": {
        "type": [
          "string",
          "null"
        ],
        "description": "Image URL - e.g. LinkedIn profile picture or the company logo. Do not pass anything here if the object is not a contact or company (it will be ignored)."
      },
      "restrictedFieldNames": {
        "type": "array",
        "items": {
          "type": "string",
          "minLength": 1
        },
        "description": "Names of restricted fields requested by the task that were intentionally omitted. Never include their values."
      }
    },
    "additionalProperties": false
  },
  "annotations": {
    "title": "Create CRM object",
    "readOnlyHint": false,
    "openWorldHint": false,
    "destructiveHint": false
  }
}

update_objectTitles 96Descriptions 75Inputs 69Annotations 75Outputs 072/100

Title96/100

Update CRM object

The title clearly communicates the tool’s action and subject. Keep it concise.

Description75/100

Update an existing object in the system. objectSlug and id are required. If the task requested a restricted field, omit its value and list only its field name in restrictedFieldNames; permitted fields will still be updated.

Explain mutations, irreversible effects, or prerequisites that matter before calling this tool.

Inputs69/100

Remove validation rules already expressed in this field’s schema; describe its meaning and usage instead.

objectSlug

Object type slug, such as contacts, companies, or deals.

Clear enough to supply this argument.

ID of the object to update.

Clear enough to supply this argument.

fields › items › name

Field name.

Remove validation rules already expressed in this field’s schema; describe its meaning and usage instead.

fields › items › value

New field value.

Explain the meaning of this value, not just its name.

restrictedFieldNames

Names of restricted fields requested by the task that were intentionally omitted. Never include their values.

Clear enough to supply this argument.

profileImageUrl

Image URL - e.g. LinkedIn profile picture or the company logo. Do not pass anything here if you are not changing the image, or if object is not a contact or company (it will be ignored).

Remove validation rules already expressed in this field’s schema; describe its meaning and usage instead.

Annotations75/100

Declare idempotentHint based on the tool’s actual behavior.

{
  "title": "Update CRM object",
  "readOnlyHint": false,
  "openWorldHint": false,
  "destructiveHint": true
}

Output schema0/100

No output schema is declared. Add one for structured results so clients can interpret returned fields.

Not declared

Original tool metadata

{
  "name": "update_object",
  "title": "Update CRM object",
  "description": "Update an existing object in the system. objectSlug and id are required. If the task requested a restricted field, omit its value and list only its field name in restrictedFieldNames; permitted fields will still be updated.",
  "inputSchema": {
    "type": "object",
    "$schema": "http://json-schema.org/draft-07/schema#",
    "required": [
      "objectSlug",
      "id",
      "fields"
    ],
    "properties": {
      "id": {
        "type": "number",
        "description": "ID of the object to update."
      },
      "fields": {
        "type": "array",
        "items": {
          "type": "object",
          "required": [
            "name"
          ],
          "properties": {
            "name": {
              "type": "string",
              "minLength": 1,
              "description": "Field name."
            },
            "value": {
              "anyOf": [
                {
                  "anyOf": [
                    {
                      "not": {}
                    },
                    {}
                  ]
                },
                {
                  "type": "null"
                }
              ],
              "description": "New field value."
            }
          },
          "additionalProperties": false
        },
        "description": "Field values to change, as name/value pairs."
      },
      "objectSlug": {
        "type": "string",
        "minLength": 1,
        "description": "Object type slug, such as contacts, companies, or deals."
      },
      "profileImageUrl": {
        "anyOf": [
          {
            "anyOf": [
              {
                "not": {}
              },
              {
                "type": "string",
                "description": "Image URL - e.g. LinkedIn profile picture or the company logo. Do not pass anything here if you are not changing the image, or if object is not a contact or company (it will be ignored)."
              }
            ],
            "description": "Image URL - e.g. LinkedIn profile picture or the company logo. Do not pass anything here if you are not changing the image, or if object is not a contact or company (it will be ignored)."
          },
          {
            "type": "null"
          }
        ],
        "description": "Image URL - e.g. LinkedIn profile picture or the company logo. Do not pass anything here if you are not changing the image, or if object is not a contact or company (it will be ignored)."
      },
      "restrictedFieldNames": {
        "type": "array",
        "items": {
          "type": "string",
          "minLength": 1
        },
        "description": "Names of restricted fields requested by the task that were intentionally omitted. Never include their values."
      }
    },
    "additionalProperties": false
  },
  "annotations": {
    "title": "Update CRM object",
    "readOnlyHint": false,
    "openWorldHint": false,
    "destructiveHint": true
  }
}

delete_objectTitles 99Descriptions 69Inputs 75Annotations 75Outputs 072/100

Title99/100

Delete CRM object

The title clearly communicates the tool’s action and subject. Keep it concise.

Description69/100

Delete an existing object in the system. id and objectSlug are required.

Explain mutations, irreversible effects, or prerequisites that matter before calling this tool.

Inputs75/100

Clarify units, where to obtain this value, or conventions not already expressed in the schema.

objectSlug

Object type slug, such as contacts, companies, or deals.

Clear enough to supply this argument.

ID of the object to delete.

Clarify units, where to obtain this value, or conventions not already expressed in the schema.

_skip_approval

Task runs only: set true to auto-execute without approval when this action is routine and low-risk.

Explain what happens when this optional value is omitted.

Annotations75/100

Declare idempotentHint based on the tool’s actual behavior.

{
  "title": "Delete CRM object",
  "readOnlyHint": false,
  "openWorldHint": false,
  "destructiveHint": true
}

Output schema0/100

No output schema is declared. Add one for structured results so clients can interpret returned fields.

Not declared

Original tool metadata

{
  "name": "delete_object",
  "title": "Delete CRM object",
  "description": "Delete an existing object in the system. id and objectSlug are required.",
  "inputSchema": {
    "type": "object",
    "$schema": "http://json-schema.org/draft-07/schema#",
    "required": [
      "objectSlug",
      "id"
    ],
    "properties": {
      "id": {
        "type": "number",
        "description": "ID of the object to delete."
      },
      "objectSlug": {
        "type": "string",
        "minLength": 1,
        "description": "Object type slug, such as contacts, companies, or deals."
      },
      "_skip_approval": {
        "type": "boolean",
        "description": "Task runs only: set true to auto-execute without approval when this action is routine and low-risk."
      }
    },
    "additionalProperties": false
  },
  "annotations": {
    "title": "Delete CRM object",
    "readOnlyHint": false,
    "openWorldHint": false,
    "destructiveHint": true
  }
}

list_viewsTitles 99Descriptions 85Inputs 90Annotations 75Outputs 081/100

Title99/100

List saved views

The title clearly communicates the tool’s action and subject. Keep it concise.

Description85/100

List all saved views for an object type. Views are pre-configured filters, columns, and sorting that users have saved. Use this to discover available views before executing them with execute_view.

Describe the returned information so the agent knows whether it will answer the request.

Inputs90/100

Input documentation gives an agent enough context to populate the arguments.

object_slug

The object type slug (e.g., "contacts", "companies", "deals")

Clear enough to supply this argument.

Annotations75/100

Declare idempotentHint based on the tool’s actual behavior.

{
  "title": "List saved views",
  "readOnlyHint": true,
  "openWorldHint": false,
  "destructiveHint": false
}

Output schema0/100

No output schema is declared. Add one for structured results so clients can interpret returned fields.

Not declared

Original tool metadata

{
  "name": "list_views",
  "title": "List saved views",
  "description": "List all saved views for an object type. Views are pre-configured filters, columns, and sorting that users have saved. Use this to discover available views before executing them with execute_view.",
  "inputSchema": {
    "type": "object",
    "$schema": "http://json-schema.org/draft-07/schema#",
    "required": [
      "object_slug"
    ],
    "properties": {
      "object_slug": {
        "type": "string",
        "description": "The object type slug (e.g., \"contacts\", \"companies\", \"deals\")"
      }
    },
    "additionalProperties": false
  },
  "annotations": {
    "title": "List saved views",
    "readOnlyHint": true,
    "openWorldHint": false,
    "destructiveHint": false
  }
}

list_listsTitles 86Descriptions 71Inputs 96Annotations 75Outputs 075/100

Title86/100

List lists

The title clearly communicates the tool’s action and subject. Keep it concise.

Description71/100

Browse available lists for the organization. Supports optional search by name and filtering by object type slug.

Describe the returned information so the agent knows whether it will answer the request.

Inputs96/100

Input documentation gives an agent enough context to populate the arguments.

Filter lists by name (case-insensitive partial match)

Clear enough to supply this argument.

objectType

Filter by object type slug (e.g., "contacts", "companies", or a custom type slug). Resolves slug to object_type_id via object_types table.

Clear enough to supply this argument.

Annotations75/100

Declare idempotentHint based on the tool’s actual behavior.

{
  "title": "List lists",
  "readOnlyHint": true,
  "openWorldHint": false,
  "destructiveHint": false
}

Output schema0/100

No output schema is declared. Add one for structured results so clients can interpret returned fields.

Not declared

Original tool metadata

{
  "name": "list_lists",
  "title": "List lists",
  "description": "Browse available lists for the organization. Supports optional search by name and filtering by object type slug.",
  "inputSchema": {
    "type": "object",
    "$schema": "http://json-schema.org/draft-07/schema#",
    "properties": {
      "search": {
        "type": "string",
        "description": "Filter lists by name (case-insensitive partial match)"
      },
      "objectType": {
        "type": "string",
        "description": "Filter by object type slug (e.g., \"contacts\", \"companies\", or a custom type slug). Resolves slug to object_type_id via object_types table."
      }
    },
    "additionalProperties": false
  },
  "annotations": {
    "title": "List lists",
    "readOnlyHint": true,
    "openWorldHint": false,
    "destructiveHint": false
  }
}

get_listTitles 99Descriptions 92Inputs 83Annotations 75Outputs 081/100

Title99/100

Get list members

The title clearly communicates the tool’s action and subject. Keep it concise.

Description92/100

Get details for one Item list along with its member records, including the full field data for each member. Supports limit and offset for paginating large lists. List IDs come from list_lists.

The description explains the capability, when to use it, and what to expect.

Inputs83/100

Clarify units, where to obtain this value, or conventions not already expressed in the schema.

listId

The ID of the list to retrieve

Clarify units, where to obtain this value, or conventions not already expressed in the schema.

limit

Max number of members to return (default: 1000). Use for pagination.

Clear enough to supply this argument.

offset

Offset for pagination (default: 0)

Clear enough to supply this argument.

Annotations75/100

Declare idempotentHint based on the tool’s actual behavior.

{
  "title": "Get list members",
  "readOnlyHint": true,
  "openWorldHint": false,
  "destructiveHint": false
}

Output schema0/100

No output schema is declared. Add one for structured results so clients can interpret returned fields.

Not declared

Original tool metadata

{
  "name": "get_list",
  "title": "Get list members",
  "description": "Get details for one Item list along with its member records, including the full field data for each member. Supports limit and offset for paginating large lists. List IDs come from list_lists.",
  "inputSchema": {
    "type": "object",
    "$schema": "http://json-schema.org/draft-07/schema#",
    "required": [
      "listId"
    ],
    "properties": {
      "limit": {
        "type": "number",
        "description": "Max number of members to return (default: 1000). Use for pagination."
      },
      "listId": {
        "type": "number",
        "description": "The ID of the list to retrieve"
      },
      "offset": {
        "type": "number",
        "description": "Offset for pagination (default: 0)"
      }
    },
    "additionalProperties": false
  },
  "annotations": {
    "title": "Get list members",
    "readOnlyHint": true,
    "openWorldHint": false,
    "destructiveHint": false
  }
}

get_last_phone_callTitles 100Descriptions 97Inputs 75Annotations 75Outputs 080/100

Title100/100

Get last phone call

The title clearly communicates the tool’s action and subject. Keep it concise.

Description97/100

Read the latest Item-native phone call linked to a Contact or Company, including its timestamp, title, and description. Pass exactly one positive contactId or companyId; omit the unused ID or pass null. Access follows workspace membership and Last Phone Call field visibility.

The description explains the capability, when to use it, and what to expect.

Inputs75/100

Remove validation rules already expressed in this field’s schema; describe its meaning and usage instead.

contactId

Contact ID whose exact Last Phone Call is requested. Pass null when using companyId.

Remove validation rules already expressed in this field’s schema; describe its meaning and usage instead.

companyId

Company ID whose exact Last Phone Call is requested. Pass null when using contactId.

Remove validation rules already expressed in this field’s schema; describe its meaning and usage instead.

Annotations75/100

Declare idempotentHint based on the tool’s actual behavior.

{
  "title": "Get last phone call",
  "readOnlyHint": true,
  "openWorldHint": false,
  "destructiveHint": false
}

Output schema0/100

No output schema is declared. Add one for structured results so clients can interpret returned fields.

Not declared

Original tool metadata

{
  "name": "get_last_phone_call",
  "title": "Get last phone call",
  "description": "Read the latest Item-native phone call linked to a Contact or Company, including its timestamp, title, and description. Pass exactly one positive contactId or companyId; omit the unused ID or pass null. Access follows workspace membership and Last Phone Call field visibility.",
  "inputSchema": {
    "type": "object",
    "$schema": "http://json-schema.org/draft-07/schema#",
    "properties": {
      "companyId": {
        "anyOf": [
          {
            "type": "integer",
            "exclusiveMinimum": 0
          },
          {
            "type": "null"
          }
        ],
        "description": "Company ID whose exact Last Phone Call is requested. Pass null when using contactId."
      },
      "contactId": {
        "anyOf": [
          {
            "type": "integer",
            "exclusiveMinimum": 0
          },
          {
            "type": "null"
          }
        ],
        "description": "Contact ID whose exact Last Phone Call is requested. Pass null when using companyId."
      }
    },
    "additionalProperties": false
  },
  "annotations": {
    "title": "Get last phone call",
    "readOnlyHint": true,
    "openWorldHint": false,
    "destructiveHint": false
  }
}

create_listTitles 98Descriptions 81Inputs 81Annotations 75Outputs 077/100

Title98/100

Create list

The title clearly communicates the tool’s action and subject. Keep it concise.

Description81/100

Create a new list for organizing Item objects of a single object type. Requires a name and the object type slug, with an optional description. Use manage_list for membership changes; Item CLI sessions are available only in the web Assistant.

The description explains the capability, when to use it, and what to expect.

Inputs81/100

Remove validation rules already expressed in this field’s schema; describe its meaning and usage instead.

name

Name for the new list

Clear enough to supply this argument.

objectType

Object type slug (e.g., "contacts", "companies", or a custom type slug)

Clear enough to supply this argument.

description

Optional description for the list

Remove validation rules already expressed in this field’s schema; describe its meaning and usage instead.

Annotations75/100

Declare idempotentHint based on the tool’s actual behavior.

{
  "title": "Create list",
  "readOnlyHint": false,
  "openWorldHint": false,
  "destructiveHint": false
}

Output schema0/100

No output schema is declared. Add one for structured results so clients can interpret returned fields.

Not declared

Original tool metadata

{
  "name": "create_list",
  "title": "Create list",
  "description": "Create a new list for organizing Item objects of a single object type. Requires a name and the object type slug, with an optional description. Use manage_list for membership changes; Item CLI sessions are available only in the web Assistant.",
  "inputSchema": {
    "type": "object",
    "$schema": "http://json-schema.org/draft-07/schema#",
    "required": [
      "name",
      "objectType"
    ],
    "properties": {
      "name": {
        "type": "string",
        "description": "Name for the new list"
      },
      "objectType": {
        "type": "string",
        "description": "Object type slug (e.g., \"contacts\", \"companies\", or a custom type slug)"
      },
      "description": {
        "type": "string",
        "description": "Optional description for the list"
      }
    },
    "additionalProperties": false
  },
  "annotations": {
    "title": "Create list",
    "readOnlyHint": false,
    "openWorldHint": false,
    "destructiveHint": false
  }
}

create_noteTitles 99Descriptions 56Inputs 50Annotations 75Outputs 062/100

Title99/100

Create note

The title clearly communicates the tool’s action and subject. Keep it concise.

Description56/100

Create a new note

Describe the concrete operation and its scope instead of restating the tool name.

Inputs50/100

Explain the meaning of this value, not just its name.

content

The content of the note

Explain the meaning of this value, not just its name.

title

The title of the note

Explain the meaning of this value, not just its name.

object_slug

The slug of the object the note is associated with

Clarify units, where to obtain this value, or conventions not already expressed in the schema.

owned_by_object_id

The object id to create the note for

Clarify units, where to obtain this value, or conventions not already expressed in the schema.

Annotations75/100

Declare idempotentHint based on the tool’s actual behavior.

{
  "title": "Create note",
  "readOnlyHint": false,
  "openWorldHint": false,
  "destructiveHint": false
}

Output schema0/100

No output schema is declared. Add one for structured results so clients can interpret returned fields.

Not declared

Original tool metadata

{
  "name": "create_note",
  "title": "Create note",
  "description": "Create a new note",
  "inputSchema": {
    "type": "object",
    "$schema": "http://json-schema.org/draft-07/schema#",
    "required": [
      "content",
      "title",
      "object_slug"
    ],
    "properties": {
      "title": {
        "type": [
          "string",
          "null"
        ],
        "description": "The title of the note"
      },
      "content": {
        "type": [
          "string",
          "null"
        ],
        "description": "The content of the note"
      },
      "object_slug": {
        "type": "string",
        "description": "The slug of the object the note is associated with"
      },
      "owned_by_object_id": {
        "type": "number",
        "description": "The object id to create the note for"
      }
    },
    "additionalProperties": false
  },
  "annotations": {
    "title": "Create note",
    "readOnlyHint": false,
    "openWorldHint": false,
    "destructiveHint": false
  }
}

edit_noteTitles 100Descriptions 78Inputs 76Annotations 75Outputs 075/100

Title100/100

Edit note

The title clearly communicates the tool’s action and subject. Keep it concise.

Description78/100

Edit an existing note. For title or content, provide both the exact value you read and its replacement; a newer human edit returns a conflict.

Explain mutations, irreversible effects, or prerequisites that matter before calling this tool.

Inputs76/100

Clarify units, where to obtain this value, or conventions not already expressed in the schema.

content › expected

The exact current value read before generating this edit

Clear enough to supply this argument.

content › next

The replacement value

Clear enough to supply this argument.

title › expected

The exact current value read before generating this edit

Clear enough to supply this argument.

title › next

The replacement value

Clear enough to supply this argument.

emoji

The emoji of the note in unified code

Clarify units, where to obtain this value, or conventions not already expressed in the schema.

object_slug

The slug of the object the note is associated with

Clarify units, where to obtain this value, or conventions not already expressed in the schema.

owned_by_object_id

The id of the object to update the note for

Clarify units, where to obtain this value, or conventions not already expressed in the schema.

The id of the note to update

Clear enough to supply this argument.

Annotations75/100

Declare idempotentHint based on the tool’s actual behavior.

{
  "title": "Edit note",
  "readOnlyHint": false,
  "openWorldHint": false,
  "destructiveHint": true
}

Output schema0/100

No output schema is declared. Add one for structured results so clients can interpret returned fields.

Not declared

Original tool metadata

{
  "name": "edit_note",
  "title": "Edit note",
  "description": "Edit an existing note. For title or content, provide both the exact value you read and its replacement; a newer human edit returns a conflict.",
  "inputSchema": {
    "type": "object",
    "$schema": "http://json-schema.org/draft-07/schema#",
    "required": [
      "id"
    ],
    "properties": {
      "id": {
        "type": "string",
        "description": "The id of the note to update"
      },
      "emoji": {
        "type": [
          "string",
          "null"
        ],
        "description": "The emoji of the note in unified code"
      },
      "title": {
        "type": "object",
        "required": [
          "expected",
          "next"
        ],
        "properties": {
          "next": {
            "$ref": "#/properties/content/properties/next"
          },
          "expected": {
            "$ref": "#/properties/content/properties/expected"
          }
        },
        "description": "Expected and replacement note title",
        "additionalProperties": false
      },
      "content": {
        "type": "object",
        "required": [
          "expected",
          "next"
        ],
        "properties": {
          "next": {
            "type": [
              "string",
              "null"
            ],
            "description": "The replacement value"
          },
          "expected": {
            "type": [
              "string",
              "null"
            ],
            "description": "The exact current value read before generating this edit"
          }
        },
        "description": "Expected and replacement note content",
        "additionalProperties": false
      },
      "object_slug": {
        "type": "string",
        "description": "The slug of the object the note is associated with"
      },
      "owned_by_object_id": {
        "type": "number",
        "description": "The id of the object to update the note for"
      }
    },
    "additionalProperties": false
  },
  "annotations": {
    "title": "Edit note",
    "readOnlyHint": false,
    "openWorldHint": false,
    "destructiveHint": true
  }
}

search_notesTitles 100Descriptions 50Inputs 56Annotations 75Outputs 062/100

Title100/100

Search notes

The title clearly communicates the tool’s action and subject. Keep it concise.

Description50/100

Search for notes

Describe the concrete operation and its scope instead of restating the tool name.

Inputs56/100

Clarify units, where to obtain this value, or conventions not already expressed in the schema.

query

The query in natural language to search for notes

Clear enough to supply this argument.

owned_by_object_id

The id of the object that owns the note

Clarify units, where to obtain this value, or conventions not already expressed in the schema.

object_slug

Limit search to a specific object slug

Explain the meaning of this value, not just its name.

date

The date to search for notes on

Clarify units, where to obtain this value, or conventions not already expressed in the schema.

created_by_user_id

The id of the user that created the note

Clarify units, where to obtain this value, or conventions not already expressed in the schema.

last_edited_by_user_id

The id of the user that last edited the note

Clarify units, where to obtain this value, or conventions not already expressed in the schema.

from_date

The date to search for notes from

Clarify units, where to obtain this value, or conventions not already expressed in the schema.

to_date

The date to search for notes to

Clarify units, where to obtain this value, or conventions not already expressed in the schema.

Annotations75/100

Declare idempotentHint based on the tool’s actual behavior.

{
  "title": "Search notes",
  "readOnlyHint": true,
  "openWorldHint": false,
  "destructiveHint": false
}

Output schema0/100

No output schema is declared. Add one for structured results so clients can interpret returned fields.

Not declared

Original tool metadata

{
  "name": "search_notes",
  "title": "Search notes",
  "description": "Search for notes",
  "inputSchema": {
    "type": "object",
    "$schema": "http://json-schema.org/draft-07/schema#",
    "required": [
      "query"
    ],
    "properties": {
      "date": {
        "type": "string",
        "description": "The date to search for notes on"
      },
      "query": {
        "type": "string",
        "description": "The query in natural language to search for notes"
      },
      "to_date": {
        "type": "string",
        "description": "The date to search for notes to"
      },
      "from_date": {
        "type": "string",
        "description": "The date to search for notes from"
      },
      "object_slug": {
        "type": "string",
        "description": "Limit search to a specific object slug"
      },
      "created_by_user_id": {
        "type": "string",
        "description": "The id of the user that created the note"
      },
      "owned_by_object_id": {
        "type": "number",
        "description": "The id of the object that owns the note"
      },
      "last_edited_by_user_id": {
        "type": "string",
        "description": "The id of the user that last edited the note"
      }
    },
    "additionalProperties": false
  },
  "annotations": {
    "title": "Search notes",
    "readOnlyHint": true,
    "openWorldHint": false,
    "destructiveHint": false
  }
}

search_tasksTitles 100Descriptions 97Inputs 92Annotations 75Outputs 085/100

Title100/100

Search tasks

The title clearly communicates the tool’s action and subject. Keep it concise.

Description97/100

Search the user's tasks by status, scheduled/created/completed dates, assignee, associated objects (contact, company, custom object), and an optional free-text `query` string. Internal lookup tool — the result is consumed by the model only. This tool DOES NOT render a card. Do NOT summarize the result back to the user as a list, table, or bulleted enumeration — the search is a programmatic lookup, not a user-facing display. When to use: - Use `search_tasks` whenever you need to look up a task to act on — e.g. resolving a `task_id` for `update_task`, `delete_task`, or to re-open a terminal human task. - Use `list_tasks` (not this tool) when the user explicitly wants to see their tasks ("show me my tasks", "what's on my plate?"). `list_tasks` renders a rich task-list card; `search_tasks` does not. Input shape is identical to `list_tasks`. The same filters apply: - `query`: free-text search across the task title and planned steps. Whitespace-separated terms are AND-ed together; each term must appear in either the title or the planned steps. Use when the user names a task by content (e.g. "the task about Rosa", "the enrichment task"), and pass only the CONTENT words — never record-type words like "follow-up", "task", "reminder" or "to-do", which exclude every matching task whose title happens not to contain them. Pass null when the user identifies the task by who it involves rather than by content. - `status`: one of the user-facing labels (`Pending`, `Scheduled`, `In-progress`, `Awaiting Trigger`, `Input Needed`, `Pending Approval`, `Completed`, `Failed`, `Canceled`), or `null` to search ALL statuses. Pass null whenever the user hasn't named a state. Same for `assignee_type`. - Date ranges (`scheduled_after` / `scheduled_before`, `created_after` / `created_before`, `completed_after` / `completed_before`), assignee filters (`assignee_type`, `assignee_user_id`, `created_by_user_id`), and association filters (`contact_id`, `contact_company_id`, `custom_object_id`) all behave the same way as in `list_tasks`. `scheduled_after`/`scheduled_before` EXCLUDE undated tasks — omit them unless the user asked about a time window. `contact_company_id` matches both tasks linked directly to the company and tasks linked to one of its contacts. - `query` may be combined with metadata filters to narrow results (e.g. `query: "Rosa"` together with `created_after` / `created_before` covering yesterday). Examples: - "Push the prep call task to Friday" → first call `search_tasks { query: "prep call" }` to resolve the task id, then `update_task { task_id, scheduled_for: '...' }`. - "Cancel the follow-up I asked you to do for Rosa" → `search_tasks { query: "Rosa" }` — "follow-up" is a record type, not content, so it does NOT belong in `query` — then `update_task { task_id, status: 'cancelled' }`. - "Reassign the Acme outreach task to Bob" → `search_tasks { query: "Acme outreach" }` (also resolve Bob's user_id via `internal_search`), then `update_task { task_id, assignee_user_id }`.

The description explains the capability, when to use it, and what to expect.

Inputs92/100

Remove validation rules already expressed in this field’s schema; describe its meaning and usage instead.

status

Pass null to search ALL statuses — do that for open-ended questions like "do I have any tasks with Acme?", where guessing a label silently hides matches. Only pass a label when the user names a state. Filter by user-facing status label (the values shown in the UI). "Pending" = scheduled human task. "Scheduled" = scheduled AI task or AI task paused mid-execution. "In-progress" = currently running. "Awaiting Trigger" = waiting for a trigger event to fire. "Input Needed" = paused waiting for the user to provide input. "Pending Approval" = paused waiting for the user to approve an action. "Completed" = succeeded. "Failed" = errored or exited without completing. "Canceled" = explicitly canceled.

Remove validation rules already expressed in this field’s schema; describe its meaning and usage instead.

scheduled_after

ISO 8601 datetime - only return tasks scheduled after this time. WARNING: excludes undated tasks (no scheduled date), which are common. Omit unless the user asked about a specific time window.

Clear enough to supply this argument.

scheduled_before

ISO 8601 datetime - only return tasks scheduled before this time. WARNING: excludes undated tasks (no scheduled date), which are common. Omit unless the user asked about a specific time window.

Clear enough to supply this argument.

assignee_type

Filter by assignee type. Pass null to search both AI and human tasks — do that unless the user asked for one kind specifically.

Clear enough to supply this argument.

contact_id

Filter by associated contact

Explain the meaning of this value, not just its name.

contact_company_id

Filter by associated company. Matches tasks linked directly to the company AND tasks linked to a contact who works there.

Clear enough to supply this argument.

custom_object_id

Filter by associated custom object

Explain the meaning of this value, not just its name.

limit

Maximum number of tasks to return (default: 20)

Clear enough to supply this argument.

query

Free-text search across the task title and planned steps. Use when the user names a task by content or phrase (e.g. "the task about Rosa", "enriching profiles"). Whitespace-separated terms are AND-ed together; each term must appear in either the title or the planned steps. Pass null for pure metadata queries (e.g. "my running tasks"). NEVER put generic record-type words here — "follow-up", "task", "reminder", "to-do", "meeting", "action item" describe the KIND of record, not its content, and will exclude every matching task whose title happens not to contain that word. When the user asks "any follow-ups/tasks with <person or company>?", resolve the person or company and filter by contact_id / contact_company_id with query: null.

Clear enough to supply this argument.

assignee_user_id

Filter by the user the task is assigned to (UUID of the user profile).

Clear enough to supply this argument.

created_by_user_id

Filter by the user who created the task (UUID of the user profile).

Clear enough to supply this argument.

created_after

ISO 8601 datetime — only return tasks created at or after this time.

Clear enough to supply this argument.

created_before

ISO 8601 datetime — only return tasks created at or before this time.

Clear enough to supply this argument.

completed_after

ISO 8601 datetime — only return tasks completed at or after this time.

Clear enough to supply this argument.

completed_before

ISO 8601 datetime — only return tasks completed at or before this time.

Clear enough to supply this argument.

Annotations75/100

Declare idempotentHint based on the tool’s actual behavior.

{
  "title": "Search tasks",
  "readOnlyHint": true,
  "openWorldHint": false,
  "destructiveHint": false
}

Output schema0/100

No output schema is declared. Add one for structured results so clients can interpret returned fields.

Not declared

Original tool metadata

{
  "name": "search_tasks",
  "title": "Search tasks",
  "description": "Search the user's tasks by status, scheduled/created/completed dates, assignee, associated objects (contact, company, custom object), and an optional free-text `query` string. Internal lookup tool — the result is consumed by the model only.\n\nThis tool DOES NOT render a card. Do NOT summarize the result back to the user as a list, table, or bulleted enumeration — the search is a programmatic lookup, not a user-facing display.\n\nWhen to use:\n- Use `search_tasks` whenever you need to look up a task to act on — e.g. resolving a `task_id` for `update_task`, `delete_task`, or to re-open a terminal human task.\n- Use `list_tasks` (not this tool) when the user explicitly wants to see their tasks (\"show me my tasks\", \"what's on my plate?\"). `list_tasks` renders a rich task-list card; `search_tasks` does not.\n\nInput shape is identical to `list_tasks`. The same filters apply:\n- `query`: free-text search across the task title and planned steps. Whitespace-separated terms are AND-ed together; each term must appear in either the title or the planned steps. Use when the user names a task by content (e.g. \"the task about Rosa\", \"the enrichment task\"), and pass only the CONTENT words — never record-type words like \"follow-up\", \"task\", \"reminder\" or \"to-do\", which exclude every matching task whose title happens not to contain them. Pass null when the user identifies the task by who it involves rather than by content.\n- `status`: one of the user-facing labels (`Pending`, `Scheduled`, `In-progress`, `Awaiting Trigger`, `Input Needed`, `Pending Approval`, `Completed`, `Failed`, `Canceled`), or `null` to search ALL statuses. Pass null whenever the user hasn't named a state. Same for `assignee_type`.\n- Date ranges (`scheduled_after` / `scheduled_before`, `created_after` / `created_before`, `completed_after` / `completed_before`), assignee filters (`assignee_type`, `assignee_user_id`, `created_by_user_id`), and association filters (`contact_id`, `contact_company_id`, `custom_object_id`) all behave the same way as in `list_tasks`. `scheduled_after`/`scheduled_before` EXCLUDE undated tasks — omit them unless the user asked about a time window. `contact_company_id` matches both tasks linked directly to the company and tasks linked to one of its contacts.\n- `query` may be combined with metadata filters to narrow results (e.g. `query: \"Rosa\"` together with `created_after` / `created_before` covering yesterday).\n\nExamples:\n- \"Push the prep call task to Friday\" → first call `search_tasks { query: \"prep call\" }` to resolve the task id, then `update_task { task_id, scheduled_for: '...' }`.\n- \"Cancel the follow-up I asked you to do for Rosa\" → `search_tasks { query: \"Rosa\" }` — \"follow-up\" is a record type, not content, so it does NOT belong in `query` — then `update_task { task_id, status: 'cancelled' }`.\n- \"Reassign the Acme outreach task to Bob\" → `search_tasks { query: \"Acme outreach\" }` (also resolve Bob's user_id via `internal_search`), then `update_task { task_id, assignee_user_id }`.",
  "inputSchema": {
    "type": "object",
    "$schema": "http://json-schema.org/draft-07/schema#",
    "properties": {
      "limit": {
        "type": "number",
        "default": 20,
        "description": "Maximum number of tasks to return (default: 20)"
      },
      "query": {
        "type": [
          "string",
          "null"
        ],
        "description": "Free-text search across the task title and planned steps. Use when the user names a task by content or phrase (e.g. \"the task about Rosa\", \"enriching profiles\"). Whitespace-separated terms are AND-ed together; each term must appear in either the title or the planned steps. Pass null for pure metadata queries (e.g. \"my running tasks\"). NEVER put generic record-type words here — \"follow-up\", \"task\", \"reminder\", \"to-do\", \"meeting\", \"action item\" describe the KIND of record, not its content, and will exclude every matching task whose title happens not to contain that word. When the user asks \"any follow-ups/tasks with <person or company>?\", resolve the person or company and filter by contact_id / contact_company_id with query: null."
      },
      "status": {
        "anyOf": [
          {
            "enum": [
              "Pending",
              "Scheduled",
              "In-progress",
              "Awaiting Trigger",
              "Input Needed",
              "Pending Approval",
              "Completed",
              "Failed",
              "Canceled"
            ],
            "type": "string"
          },
          {
            "type": "null"
          }
        ],
        "description": "Pass null to search ALL statuses — do that for open-ended questions like \"do I have any tasks with Acme?\", where guessing a label silently hides matches. Only pass a label when the user names a state. Filter by user-facing status label (the values shown in the UI). \"Pending\" = scheduled human task. \"Scheduled\" = scheduled AI task or AI task paused mid-execution. \"In-progress\" = currently running. \"Awaiting Trigger\" = waiting for a trigger event to fire. \"Input Needed\" = paused waiting for the user to provide input. \"Pending Approval\" = paused waiting for the user to approve an action. \"Completed\" = succeeded. \"Failed\" = errored or exited without completing. \"Canceled\" = explicitly canceled."
      },
      "contact_id": {
        "type": "number",
        "description": "Filter by associated contact"
      },
      "assignee_type": {
        "anyOf": [
          {
            "enum": [
              "ai",
              "user"
            ],
            "type": "string"
          },
          {
            "type": "null"
          }
        ],
        "description": "Filter by assignee type. Pass null to search both AI and human tasks — do that unless the user asked for one kind specifically."
      },
      "created_after": {
        "type": "string",
        "description": "ISO 8601 datetime — only return tasks created at or after this time."
      },
      "created_before": {
        "type": "string",
        "description": "ISO 8601 datetime — only return tasks created at or before this time."
      },
      "completed_after": {
        "type": "string",
        "description": "ISO 8601 datetime — only return tasks completed at or after this time."
      },
      "scheduled_after": {
        "type": "string",
        "description": "ISO 8601 datetime - only return tasks scheduled after this time. WARNING: excludes undated tasks (no scheduled date), which are common. Omit unless the user asked about a specific time window."
      },
      "assignee_user_id": {
        "type": "string",
        "description": "Filter by the user the task is assigned to (UUID of the user profile)."
      },
      "completed_before": {
        "type": "string",
        "description": "ISO 8601 datetime — only return tasks completed at or before this time."
      },
      "custom_object_id": {
        "type": "number",
        "description": "Filter by associated custom object"
      },
      "scheduled_before": {
        "type": "string",
        "description": "ISO 8601 datetime - only return tasks scheduled before this time. WARNING: excludes undated tasks (no scheduled date), which are common. Omit unless the user asked about a specific time window."
      },
      "contact_company_id": {
        "type": "number",
        "description": "Filter by associated company. Matches tasks linked directly to the company AND tasks linked to a contact who works there."
      },
      "created_by_user_id": {
        "type": "string",
        "description": "Filter by the user who created the task (UUID of the user profile)."
      }
    },
    "additionalProperties": false
  },
  "annotations": {
    "title": "Search tasks",
    "readOnlyHint": true,
    "openWorldHint": false,
    "destructiveHint": false
  }
}

create_taskTitles 99Descriptions 82Inputs 86Annotations 75Outputs 079/100

Title99/100

Create task

The title clearly communicates the tool’s action and subject. Keep it concise.

Description82/100

Create a task in Item, assigned to a teammate or to the Item AI. A task runs immediately by default, at a future datetime when scheduled_for is set, or in response to an event (received email, calendar event, record change, webhook, or form submission) when a trigger is set. scheduled_for and trigger are mutually exclusive; a multi-step workflow with dependent steps belongs in a single task with one planned_steps entry per action.

The description explains the capability, when to use it, and what to expect.

Inputs86/100

Remove validation rules already expressed in this field’s schema; describe its meaning and usage instead.

prompt

The full instruction for what the task should accomplish (e.g., "Send John a follow-up email about our meeting")

Clear enough to supply this argument.

scheduled_for

ISO 8601 datetime string in UTC (ending with Z) for when the task should execute. If not provided, the task executes immediately. For relative times like "in 30 minutes", add the duration directly to the current UTC time. For absolute times like "Tuesday 9am", convert from the user's timezone to UTC using the offset in the system prompt. Mutually exclusive with trigger.

Remove validation rules already expressed in this field’s schema; describe its meaning and usage instead.

trigger › type

Type of trigger that will activate this task

Remove validation rules already expressed in this field’s schema; describe its meaning and usage instead.

is_recurring

Whether this reactive task should keep acting on every matching event. Set to true for recurring patterns ("whenever", "every time", "each week", "anytime", etc.). Set to false (default) for one-time reactive tasks ("when X happens", "as soon as X happens", etc.). Only relevant for reactive tasks (i.e. when `trigger` is provided). Set this to match what the user actually asked for — a task woken by an event that does not match its guard dismisses that event and stays armed, so a one-time task is no longer consumed by the first non-matching event.

Clear enough to supply this argument.

assignee_type

Who does the action? "ai" (default) if AI should DO something (send email, research, draft, notify, remind, etc.). "user" if only the human needs to do something, as part of their to-do list (with no automatic notification).

Clear enough to supply this argument.

contact_id

ID of the contact this task is associated with

Clarify units, where to obtain this value, or conventions not already expressed in the schema.

contact_company_id

ID of the company this task is associated with

Clarify units, where to obtain this value, or conventions not already expressed in the schema.

custom_object_id

ID of the custom object (e.g., deal) this task is associated with

Clarify units, where to obtain this value, or conventions not already expressed in the schema.

notes

Persistent context for the task. For AI tasks, these notes are included in execution context.

Clear enough to supply this argument.

explicit_approval_settings

Per-action approval preferences. Keys are action names (e.g., "reply_to_email", "send_slack_message"), values are booleans (true = require human review before executing, false = auto-execute). Only include actions the user explicitly confirmed preferences for.

Clear enough to supply this argument.

planned_steps

User-POV, imperative-voice plan describing what the AI should do — written as if the user is commanding the AI (e.g., ["Send a follow-up email to the contact", "Draft a reply using the brand tone"]). Contrast with `prompt`, which is AI-POV. One discrete step per array element, no numbering prefixes. Populate for AI tasks (typically 2–5 steps); OMIT for human tasks (assignee_type="user").

Clear enough to supply this argument.

planned_tools › anyOf › 0 › anyOf › 1 › items › name

The tool identifier (e.g., "send_email", "web_search"), chosen from the assistant's currently available tools.

Clear enough to supply this argument.

planned_tools › anyOf › 0 › anyOf › 1 › items › required

true if the task CANNOT complete without this tool — if its integration is not connected, task creation will be BLOCKED with an error naming the missing integration. false if the tool is helpful but has fallbacks/alternatives — missing integration yields a non-blocking WARNING and the tool is excluded from the persisted list, but creation still proceeds.

Clear enough to supply this argument.

assignee_user_id

Assign the task to a specific human teammate by their user_id (UUID). If omitted, the task is assigned to the calling user (default). To assign to a teammate by name (e.g., "Andres"), FIRST call internal_search to resolve the teammate name to a user_id UUID, then pass that UUID here. This field is valid for BOTH assignee_type="user" (sets the human assignee doing the work) AND assignee_type="ai" (sets the human owner/supervisor of the AI task). Org membership is enforced server-side; cross-org user IDs are rejected.

Clear enough to supply this argument.

Annotations75/100

Declare idempotentHint based on the tool’s actual behavior.

{
  "title": "Create task",
  "readOnlyHint": false,
  "openWorldHint": false,
  "destructiveHint": false
}

Output schema0/100

No output schema is declared. Add one for structured results so clients can interpret returned fields.

Not declared

Original tool metadata

{
  "name": "create_task",
  "title": "Create task",
  "description": "Create a task in Item, assigned to a teammate or to the Item AI. A task runs immediately by default, at a future datetime when scheduled_for is set, or in response to an event (received email, calendar event, record change, webhook, or form submission) when a trigger is set. scheduled_for and trigger are mutually exclusive; a multi-step workflow with dependent steps belongs in a single task with one planned_steps entry per action.",
  "inputSchema": {
    "type": "object",
    "$schema": "http://json-schema.org/draft-07/schema#",
    "required": [
      "prompt"
    ],
    "properties": {
      "notes": {
        "type": [
          "string",
          "null"
        ],
        "description": "Persistent context for the task. For AI tasks, these notes are included in execution context."
      },
      "prompt": {
        "type": "string",
        "description": "The full instruction for what the task should accomplish (e.g., \"Send John a follow-up email about our meeting\")"
      },
      "trigger": {
        "type": "object",
        "required": [
          "type"
        ],
        "properties": {
          "type": {
            "enum": [
              "new_email",
              "time_based",
              "new_calendar_event",
              "form_submission",
              "webhook",
              "record_saved",
              "connector_event",
              "calendar_event_starting"
            ],
            "type": "string",
            "description": "Type of trigger that will activate this task"
          }
        },
        "description": "Trigger condition that will activate this task. Use for reactive tasks (e.g., \"when John emails me\", \"when deal is created\"). Mutually exclusive with scheduled_for.",
        "additionalProperties": false
      },
      "contact_id": {
        "type": "number",
        "description": "ID of the contact this task is associated with"
      },
      "is_recurring": {
        "type": "boolean",
        "default": false,
        "description": "Whether this reactive task should keep acting on every matching event. Set to true for recurring patterns (\"whenever\", \"every time\", \"each week\", \"anytime\", etc.). Set to false (default) for one-time reactive tasks (\"when X happens\", \"as soon as X happens\", etc.). Only relevant for reactive tasks (i.e. when `trigger` is provided). Set this to match what the user actually asked for — a task woken by an event that does not match its guard dismisses that event and stays armed, so a one-time task is no longer consumed by the first non-matching event."
      },
      "assignee_type": {
        "enum": [
          "ai",
          "user"
        ],
        "type": "string",
        "default": "ai",
        "description": "Who does the action? \"ai\" (default) if AI should DO something (send email, research, draft, notify, remind, etc.). \"user\" if only the human needs to do something, as part of their to-do list (with no automatic notification)."
      },
      "planned_steps": {
        "anyOf": [
          {
            "anyOf": [
              {
                "not": {}
              },
              {
                "type": "array",
                "items": {
                  "type": "string"
                }
              }
            ]
          },
          {
            "type": "null"
          }
        ],
        "description": "User-POV, imperative-voice plan describing what the AI should do — written as if the user is commanding the AI (e.g., [\"Send a follow-up email to the contact\", \"Draft a reply using the brand tone\"]). Contrast with `prompt`, which is AI-POV. One discrete step per array element, no numbering prefixes. Populate for AI tasks (typically 2–5 steps); OMIT for human tasks (assignee_type=\"user\")."
      },
      "planned_tools": {
        "anyOf": [
          {
            "anyOf": [
              {
                "not": {}
              },
              {
                "type": "array",
                "items": {
                  "type": "object",
                  "required": [
                    "name",
                    "required"
                  ],
                  "properties": {
                    "name": {
                      "type": "string",
                      "description": "The tool identifier (e.g., \"send_email\", \"web_search\"), chosen from the assistant's currently available tools."
                    },
                    "required": {
                      "type": "boolean",
                      "description": "true if the task CANNOT complete without this tool — if its integration is not connected, task creation will be BLOCKED with an error naming the missing integration. false if the tool is helpful but has fallbacks/alternatives — missing integration yields a non-blocking WARNING and the tool is excluded from the persisted list, but creation still proceeds."
                    }
                  },
                  "additionalProperties": false
                }
              }
            ]
          },
          {
            "type": "null"
          }
        ],
        "description": "Array of { name, required } objects identifying the tools the task is expected to use. `required: true` means the task fundamentally cannot complete without that tool (missing integration blocks creation). `required: false` means the tool is helpful but has alternatives (missing integration produces a warning). Populate for AI tasks; OMIT for human tasks (assignee_type=\"user\")."
      },
      "scheduled_for": {
        "type": "string",
        "description": "ISO 8601 datetime string in UTC (ending with Z) for when the task should execute. If not provided, the task executes immediately. For relative times like \"in 30 minutes\", add the duration directly to the current UTC time. For absolute times like \"Tuesday 9am\", convert from the user's timezone to UTC using the offset in the system prompt. Mutually exclusive with trigger."
      },
      "assignee_user_id": {
        "type": "string",
        "format": "uuid",
        "description": "Assign the task to a specific human teammate by their user_id (UUID). If omitted, the task is assigned to the calling user (default). To assign to a teammate by name (e.g., \"Andres\"), FIRST call internal_search to resolve the teammate name to a user_id UUID, then pass that UUID here. This field is valid for BOTH assignee_type=\"user\" (sets the human assignee doing the work) AND assignee_type=\"ai\" (sets the human owner/supervisor of the AI task). Org membership is enforced server-side; cross-org user IDs are rejected."
      },
      "custom_object_id": {
        "type": "number",
        "description": "ID of the custom object (e.g., deal) this task is associated with"
      },
      "contact_company_id": {
        "type": "number",
        "description": "ID of the company this task is associated with"
      },
      "explicit_approval_settings": {
        "type": "object",
        "description": "Per-action approval preferences. Keys are action names (e.g., \"reply_to_email\", \"send_slack_message\"), values are booleans (true = require human review before executing, false = auto-execute). Only include actions the user explicitly confirmed preferences for.",
        "additionalProperties": {
          "type": "boolean"
        }
      }
    },
    "additionalProperties": false
  },
  "annotations": {
    "title": "Create task",
    "readOnlyHint": false,
    "openWorldHint": false,
    "destructiveHint": false
  }
}

update_taskTitles 96Descriptions 77Inputs 91Annotations 75Outputs 078/100

Title96/100

Update task

The title clearly communicates the tool’s action and subject. Keep it concise.

Description77/100

Update an existing Item task. Supports marking a task succeeded or cancelled, adding completion notes, resuming a blocked or errored AI task with new context (resume: true), changing the linked contact, company, or custom object, editing the plan (planned_steps) of a non-terminal AI task, renaming the task, and rescheduling via scheduled_for. Task IDs come from search_tasks.

Explain mutations, irreversible effects, or prerequisites that matter before calling this tool.

Inputs91/100

Clarify units, where to obtain this value, or conventions not already expressed in the schema.

task_id

UUID of the task to update

Clarify units, where to obtain this value, or conventions not already expressed in the schema.

clear_fields

The ONLY way to clear or unlink a field. List the fields to clear, e.g. ["notes"] to erase the Notes or ["contact_id"] to unlink the contact. Pass null or [] when not clearing anything. Passing null in a field itself does NOT clear it; null means "leave this field unchanged".

Remove validation rules already expressed in this field’s schema; describe its meaning and usage instead.

status

Update task status: "succeeded" to mark human tasks as complete, "cancelled" to cancel any task, "scheduled" to re-open a terminal human task (succeeded or cancelled) so the user can act on it again. You can cancel AI tasks, but otherwise cannot update the status of AI tasks — as their status is automatically managed by the system. Re-opening (status="scheduled") is only supported on human tasks; AI task lifecycles are managed automatically.

Remove validation rules already expressed in this field’s schema; describe its meaning and usage instead.

assignee_user_id

Reassign the task to a concrete teammate. Pass their user_id (UUID). Pass null to leave the owner unchanged — do that unless the user explicitly asked to reassign, because reassignment must be its own call and cannot be combined with any other change. User→user only — switching between AI and human assignees is not supported.

Remove validation rules already expressed in this field’s schema; describe its meaning and usage instead.

scheduled_for

ISO 8601 datetime string to reschedule the task (e.g., "2026-05-08T15:00:00Z"). Pass null to leave the schedule unchanged; to clear the date, put "scheduled_for" in clear_fields.

Clear enough to supply this argument.

resume_context

New context for the AI to continue from when `resume` is true (e.g., "Legal approved the contract, here are the details: ..."). Ignored unless `resume` is true — to record what happened on a task, use `notes`.

Clear enough to supply this argument.

notes › anyOf › 0 › text

The Notes content. With mode "append", pass ONLY the new text — the existing Notes are preserved for you, so do not restate them.

Clear enough to supply this argument.

notes › anyOf › 0 › mode

"append" adds this text to the end of the existing Notes — use this when recording what happened on a task, which is the common case. "replace" DISCARDS the existing Notes entirely; only use it when the user explicitly asks to overwrite or rewrite them, or when your text already restates everything worth keeping.

Clear enough to supply this argument.

resume

Set to true to resume a blocked AI task with `resume_context` as new context, triggering re-execution. Only valid on AI tasks in the following statuses: awaiting_user_input, error, exited.

Clear enough to supply this argument.

contact_id

Linked contact id. Pass `null` to leave the existing link unchanged. Pass a record id to set or replace the link. To UNLINK, put "contact_id" in clear_fields — passing null does NOT unlink.

Clear enough to supply this argument.

contact_company_id

Linked company id. Pass `null` to leave the existing link unchanged. Pass a record id to set or replace the link. To UNLINK, put "contact_company_id" in clear_fields — passing null does NOT unlink.

Clear enough to supply this argument.

custom_object_id

Linked custom-object id (e.g., a deal). Pass `null` to leave the existing link unchanged. Pass a record id to set or replace the link. To UNLINK, put "custom_object_id" in clear_fields — passing null does NOT unlink.

Clear enough to supply this argument.

planned_steps

User-POV, imperative-voice plan — one discrete step per array element, written as if the user is commanding the AI (e.g., ["Research the company", "Draft a tailored intro email", "Send it"]). No numbering or bullet prefixes. Use this to commit a plan you have already settled on with the user — for example, after the user confirmed the new plan during a needs_review conversation, or when the user asks to edit a task plan in chat. Only valid on non-terminal AI tasks. Pass null to leave the existing plan untouched — do that unless the user is explicitly changing the plan.

Clear enough to supply this argument.

title

New title for the task (the short user-visible headline). ONLY set this when the user explicitly asks to rename the task, or when committing a needs_review conversation where the new linked object meaningfully changes the headline. Pass null in every other case — inventing a title here silently overwrites the real one. Cannot be empty.

Clear enough to supply this argument.

Annotations75/100

Declare idempotentHint based on the tool’s actual behavior.

{
  "title": "Update task",
  "readOnlyHint": false,
  "openWorldHint": false,
  "destructiveHint": true
}

Output schema0/100

No output schema is declared. Add one for structured results so clients can interpret returned fields.

Not declared

Original tool metadata

{
  "name": "update_task",
  "title": "Update task",
  "description": "Update an existing Item task. Supports marking a task succeeded or cancelled, adding completion notes, resuming a blocked or errored AI task with new context (resume: true), changing the linked contact, company, or custom object, editing the plan (planned_steps) of a non-terminal AI task, renaming the task, and rescheduling via scheduled_for. Task IDs come from search_tasks.",
  "inputSchema": {
    "type": "object",
    "$schema": "http://json-schema.org/draft-07/schema#",
    "required": [
      "task_id"
    ],
    "properties": {
      "notes": {
        "anyOf": [
          {
            "type": "object",
            "required": [
              "text",
              "mode"
            ],
            "properties": {
              "mode": {
                "enum": [
                  "append",
                  "replace"
                ],
                "type": "string",
                "description": "\"append\" adds this text to the end of the existing Notes — use this when recording what happened on a task, which is the common case. \"replace\" DISCARDS the existing Notes entirely; only use it when the user explicitly asks to overwrite or rewrite them, or when your text already restates everything worth keeping."
              },
              "text": {
                "type": "string",
                "description": "The Notes content. With mode \"append\", pass ONLY the new text — the existing Notes are preserved for you, so do not restate them."
              }
            },
            "additionalProperties": false
          },
          {
            "type": "null"
          }
        ],
        "description": "Persistent freeform Notes for the task, and the place to record its outcome or history (e.g., \"Went great, John loved the idea\"). Included as context in future AI runs. Pass null to leave the Notes unchanged; to erase them, put \"notes\" in clear_fields."
      },
      "title": {
        "type": [
          "string",
          "null"
        ],
        "description": "New title for the task (the short user-visible headline). ONLY set this when the user explicitly asks to rename the task, or when committing a needs_review conversation where the new linked object meaningfully changes the headline. Pass null in every other case — inventing a title here silently overwrites the real one. Cannot be empty."
      },
      "resume": {
        "type": "boolean",
        "default": false,
        "description": "Set to true to resume a blocked AI task with `resume_context` as new context, triggering re-execution. Only valid on AI tasks in the following statuses: awaiting_user_input, error, exited."
      },
      "status": {
        "anyOf": [
          {
            "enum": [
              "succeeded",
              "cancelled",
              "scheduled"
            ],
            "type": "string"
          },
          {
            "type": "null"
          }
        ],
        "description": "Update task status: \"succeeded\" to mark human tasks as complete, \"cancelled\" to cancel any task, \"scheduled\" to re-open a terminal human task (succeeded or cancelled) so the user can act on it again. You can cancel AI tasks, but otherwise cannot update the status of AI tasks — as their status is automatically managed by the system. Re-opening (status=\"scheduled\") is only supported on human tasks; AI task lifecycles are managed automatically."
      },
      "task_id": {
        "type": "string",
        "description": "UUID of the task to update"
      },
      "contact_id": {
        "type": [
          "number",
          "null"
        ],
        "description": "Linked contact id. Pass `null` to leave the existing link unchanged. Pass a record id to set or replace the link. To UNLINK, put \"contact_id\" in clear_fields — passing null does NOT unlink."
      },
      "clear_fields": {
        "anyOf": [
          {
            "type": "array",
            "items": {
              "enum": [
                "scheduled_for",
                "notes",
                "contact_id",
                "contact_company_id",
                "custom_object_id"
              ],
              "type": "string"
            }
          },
          {
            "type": "null"
          }
        ],
        "description": "The ONLY way to clear or unlink a field. List the fields to clear, e.g. [\"notes\"] to erase the Notes or [\"contact_id\"] to unlink the contact. Pass null or [] when not clearing anything. Passing null in a field itself does NOT clear it; null means \"leave this field unchanged\"."
      },
      "planned_steps": {
        "anyOf": [
          {
            "type": "array",
            "items": {
              "type": "string"
            }
          },
          {
            "type": "null"
          }
        ],
        "description": "User-POV, imperative-voice plan — one discrete step per array element, written as if the user is commanding the AI (e.g., [\"Research the company\", \"Draft a tailored intro email\", \"Send it\"]). No numbering or bullet prefixes. Use this to commit a plan you have already settled on with the user — for example, after the user confirmed the new plan during a needs_review conversation, or when the user asks to edit a task plan in chat. Only valid on non-terminal AI tasks. Pass null to leave the existing plan untouched — do that unless the user is explicitly changing the plan."
      },
      "scheduled_for": {
        "anyOf": [
          {
            "type": "string",
            "format": "date-time"
          },
          {
            "type": "null"
          }
        ],
        "description": "ISO 8601 datetime string to reschedule the task (e.g., \"2026-05-08T15:00:00Z\"). Pass null to leave the schedule unchanged; to clear the date, put \"scheduled_for\" in clear_fields."
      },
      "resume_context": {
        "type": "string",
        "description": "New context for the AI to continue from when `resume` is true (e.g., \"Legal approved the contract, here are the details: ...\"). Ignored unless `resume` is true — to record what happened on a task, use `notes`."
      },
      "assignee_user_id": {
        "anyOf": [
          {
            "type": "string",
            "format": "uuid"
          },
          {
            "type": "null"
          }
        ],
        "description": "Reassign the task to a concrete teammate. Pass their user_id (UUID). Pass null to leave the owner unchanged — do that unless the user explicitly asked to reassign, because reassignment must be its own call and cannot be combined with any other change. User→user only — switching between AI and human assignees is not supported."
      },
      "custom_object_id": {
        "type": [
          "number",
          "null"
        ],
        "description": "Linked custom-object id (e.g., a deal). Pass `null` to leave the existing link unchanged. Pass a record id to set or replace the link. To UNLINK, put \"custom_object_id\" in clear_fields — passing null does NOT unlink."
      },
      "contact_company_id": {
        "type": [
          "number",
          "null"
        ],
        "description": "Linked company id. Pass `null` to leave the existing link unchanged. Pass a record id to set or replace the link. To UNLINK, put \"contact_company_id\" in clear_fields — passing null does NOT unlink."
      }
    },
    "additionalProperties": false
  },
  "annotations": {
    "title": "Update task",
    "readOnlyHint": false,
    "openWorldHint": false,
    "destructiveHint": true
  }
}

delete_taskTitles 100Descriptions 89Inputs 83Annotations 75Outputs 080/100

Title100/100

Delete task

The title clearly communicates the tool’s action and subject. Keep it concise.

Description89/100

Delete a task. Always asks for the user's confirmation before deletion. This action cannot be undone — there is no UI to restore deleted tasks via the assistant.

The description explains the capability, when to use it, and what to expect.

Inputs83/100

Explain what happens when this optional value is omitted.

task_id

UUID of the task to delete

Clear enough to supply this argument.

_skip_approval

Task runs only: set true to auto-execute without approval when this action is routine and low-risk.

Explain what happens when this optional value is omitted.

Annotations75/100

Declare idempotentHint based on the tool’s actual behavior.

{
  "title": "Delete task",
  "readOnlyHint": false,
  "openWorldHint": false,
  "destructiveHint": true
}

Output schema0/100

No output schema is declared. Add one for structured results so clients can interpret returned fields.

Not declared

Original tool metadata

{
  "name": "delete_task",
  "title": "Delete task",
  "description": "Delete a task. Always asks for the user's confirmation before deletion. This action cannot be undone — there is no UI to restore deleted tasks via the assistant.",
  "inputSchema": {
    "type": "object",
    "$schema": "http://json-schema.org/draft-07/schema#",
    "required": [
      "task_id"
    ],
    "properties": {
      "task_id": {
        "type": "string",
        "description": "UUID of the task to delete"
      },
      "_skip_approval": {
        "type": "boolean",
        "description": "Task runs only: set true to auto-execute without approval when this action is routine and low-risk."
      }
    },
    "additionalProperties": false
  },
  "annotations": {
    "title": "Delete task",
    "readOnlyHint": false,
    "openWorldHint": false,
    "destructiveHint": true
  }
}

Put your MCP to the test

A few tasks worth trying with this server, chosen from its reviewed tools. These have not been run.

01Search CRM objectsStructured read-only search over Item CRM objects.
02List saved viewsList all saved views for an object type.
03Search notesSearch for notes
04Get CRM objectFetch one item CRM object by object_slug and object_id.
05Search tasksSearch the user's tasks by status, scheduled/created/completed dates, assignee, associated objects (contact, company, custom object), and an optional free-te…

Want to run these tasks against your MCP server with us?

Test your MCP in Cloud Book a test call

About this server

Overview

Author: item
Verification: partner
Auth: OAuth required
Category: Productivity
Works with: claude
claude-desktop
claude-code
Docs: Documentation

Tools (22)

ask_item_assistantsearch_objectsget_objectlist_object_typesget_object_schemacreate_objectupdate_objectdelete_objectlist_viewsexecute_viewlist_listsget_list+10 more

FAQ

item-brain setup, tools, quality score, publishing readiness, and alternatives.

How do I add item-brain to Claude, ChatGPT, or Cursor?

Add item-brain to Claude, ChatGPT, or Cursor by connecting its remote MCP URL in each client's MCP settings. In Claude, add a custom connector. In ChatGPT, add an MCP app. In Cursor, add an MCP server under Settings. The MCP URL is https://mcp.item.app/mcp. This server requires OAuth, so complete the login prompt after you connect.

Is item-brain MCP ready for ChatGPT and Claude?

As of October 14, 2026, item-brain scores 96/100 on Manufact's MCP publishing checklist. The scan meets ChatGPT publishing checks and meets Claude publishing checks. Passing these checks does not guarantee marketplace approval.

How clear is item-brain's MCP tool documentation?

As of October 14, 2026, item-brain scores 74/100 in Manufact's review of tool titles, descriptions, input and output schemas, and behavior annotations. The review covers 22 of 22 discovered tools. The quality score measures documentation clarity; runtime behavior and marketplace readiness require separate checks.

What tools does the item-brain MCP server provide?

item-brain lists 22 MCP tools, including ask_item_assistant, search_objects, get_object, list_object_types, get_object_schema, create_object, update_object, delete_object, list_views, execute_view, list_lists, and get_list.

What are the best alternatives to item-brain?

As of October 20, 2026, alternatives to item-brain in productivity on the Manufact MCP leaderboard include Autodesk Product Help, BoardWise, FlagStat, Keenable Web Search, and PomoDial.

Does item-brain MCP require authentication?

item-brain requires authentication. Connect it in Claude, ChatGPT, or Cursor and complete the OAuth login prompt.

How can item-brain improve its MCP documentation?

manage_list is the lowest-scoring reviewed tool at 56/100. 0 of 23 inputs have descriptions. Add descriptions to the missing inputs, including where to obtain IDs and how to supply values. Open its review for feedback and a copyable fix prompt, then rerun the documentation review after updating your tool metadata.

Who develops item-brain?

item-brain is developed by item. It is listed as partner on the Claude connector directory.

Not your MCP?

Review any public or OAuth MCP server for protocol compatibility, tool documentation, and ChatGPT and Claude publishing readiness. No account required.

Get your MCP score

Back to leaderboard

Opening your score report…

Title100/100

Search tasks

The title clearly communicates the tool’s action and subject. Keep it concise.

Description97/100

The description explains the capability, when to use it, and what to expect.

Inputs92/100

Remove validation rules already expressed in this field’s schema; describe its meaning and usage instead.

status

Remove validation rules already expressed in this field’s schema; describe its meaning and usage instead.

scheduled_after

ISO 8601 datetime - only return tasks scheduled after this time. WARNING: excludes undated tasks (no scheduled date), which are common. Omit unless the user asked about a specific time window.

Clear enough to supply this argument.

scheduled_before

ISO 8601 datetime - only return tasks scheduled before this time. WARNING: excludes undated tasks (no scheduled date), which are common. Omit unless the user asked about a specific time window.

Clear enough to supply this argument.

assignee_type

Filter by assignee type. Pass null to search both AI and human tasks — do that unless the user asked for one kind specifically.

Clear enough to supply this argument.

contact_id

Filter by associated contact

Explain the meaning of this value, not just its name.

contact_company_id

Filter by associated company. Matches tasks linked directly to the company AND tasks linked to a contact who works there.

Clear enough to supply this argument.

custom_object_id

Filter by associated custom object

Explain the meaning of this value, not just its name.

limit

Maximum number of tasks to return (default: 20)

Clear enough to supply this argument.

query

Clear enough to supply this argument.

assignee_user_id

Filter by the user the task is assigned to (UUID of the user profile).

Clear enough to supply this argument.

created_by_user_id

Filter by the user who created the task (UUID of the user profile).

Clear enough to supply this argument.

created_after

ISO 8601 datetime — only return tasks created at or after this time.

Clear enough to supply this argument.

created_before

ISO 8601 datetime — only return tasks created at or before this time.

Clear enough to supply this argument.

completed_after

ISO 8601 datetime — only return tasks completed at or after this time.

Clear enough to supply this argument.

completed_before

ISO 8601 datetime — only return tasks completed at or before this time.

Clear enough to supply this argument.

Annotations75/100

Declare idempotentHint based on the tool’s actual behavior.

{
  "title": "Search tasks",
  "readOnlyHint": true,
  "openWorldHint": false,
  "destructiveHint": false
}

Output schema0/100

No output schema is declared. Add one for structured results so clients can interpret returned fields.

Not declared

Original tool metadata

{
  "name": "search_tasks",
  "title": "Search tasks",
  "description": "Search the user's tasks by status, scheduled/created/completed dates, assignee, associated objects (contact, company, custom object), and an optional free-text `query` string. Internal lookup tool — the result is consumed by the model only.\n\nThis tool DOES NOT render a card. Do NOT summarize the result back to the user as a list, table, or bulleted enumeration — the search is a programmatic lookup, not a user-facing display.\n\nWhen to use:\n- Use `search_tasks` whenever you need to look up a task to act on — e.g. resolving a `task_id` for `update_task`, `delete_task`, or to re-open a terminal human task.\n- Use `list_tasks` (not this tool) when the user explicitly wants to see their tasks (\"show me my tasks\", \"what's on my plate?\"). `list_tasks` renders a rich task-list card; `search_tasks` does not.\n\nInput shape is identical to `list_tasks`. The same filters apply:\n- `query`: free-text search across the task title and planned steps. Whitespace-separated terms are AND-ed together; each term must appear in either the title or the planned steps. Use when the user names a task by content (e.g. \"the task about Rosa\", \"the enrichment task\"), and pass only the CONTENT words — never record-type words like \"follow-up\", \"task\", \"reminder\" or \"to-do\", which exclude every matching task whose title happens not to contain them. Pass null when the user identifies the task by who it involves rather than by content.\n- `status`: one of the user-facing labels (`Pending`, `Scheduled`, `In-progress`, `Awaiting Trigger`, `Input Needed`, `Pending Approval`, `Completed`, `Failed`, `Canceled`), or `null` to search ALL statuses. Pass null whenever the user hasn't named a state. Same for `assignee_type`.\n- Date ranges (`scheduled_after` / `scheduled_before`, `created_after` / `created_before`, `completed_after` / `completed_before`), assignee filters (`assignee_type`, `assignee_user_id`, `created_by_user_id`), and association filters (`contact_id`, `contact_company_id`, `custom_object_id`) all behave the same way as in `list_tasks`. `scheduled_after`/`scheduled_before` EXCLUDE undated tasks — omit them unless the user asked about a time window. `contact_company_id` matches both tasks linked directly to the company and tasks linked to one of its contacts.\n- `query` may be combined with metadata filters to narrow results (e.g. `query: \"Rosa\"` together with `created_after` / `created_before` covering yesterday).\n\nExamples:\n- \"Push the prep call task to Friday\" → first call `search_tasks { query: \"prep call\" }` to resolve the task id, then `update_task { task_id, scheduled_for: '...' }`.\n- \"Cancel the follow-up I asked you to do for Rosa\" → `search_tasks { query: \"Rosa\" }` — \"follow-up\" is a record type, not content, so it does NOT belong in `query` — then `update_task { task_id, status: 'cancelled' }`.\n- \"Reassign the Acme outreach task to Bob\" → `search_tasks { query: \"Acme outreach\" }` (also resolve Bob's user_id via `internal_search`), then `update_task { task_id, assignee_user_id }`.",
  "inputSchema": {
    "type": "object",
    "$schema": "http://json-schema.org/draft-07/schema#",
    "properties": {
      "limit": {
        "type": "number",
        "default": 20,
        "description": "Maximum number of tasks to return (default: 20)"
      },
      "query": {
        "type": [
          "string",
          "null"
        ],
        "description": "Free-text search across the task title and planned steps. Use when the user names a task by content or phrase (e.g. \"the task about Rosa\", \"enriching profiles\"). Whitespace-separated terms are AND-ed together; each term must appear in either the title or the planned steps. Pass null for pure metadata queries (e.g. \"my running tasks\"). NEVER put generic record-type words here — \"follow-up\", \"task\", \"reminder\", \"to-do\", \"meeting\", \"action item\" describe the KIND of record, not its content, and will exclude every matching task whose title happens not to contain that word. When the user asks \"any follow-ups/tasks with <person or company>?\", resolve the person or company and filter by contact_id / contact_company_id with query: null."
      },
      "status": {
        "anyOf": [
          {
            "enum": [
              "Pending",
              "Scheduled",
              "In-progress",
              "Awaiting Trigger",
              "Input Needed",
              "Pending Approval",
              "Completed",
              "Failed",
              "Canceled"
            ],
            "type": "string"
          },
          {
            "type": "null"
          }
        ],
        "description": "Pass null to search ALL statuses — do that for open-ended questions like \"do I have any tasks with Acme?\", where guessing a label silently hides matches. Only pass a label when the user names a state. Filter by user-facing status label (the values shown in the UI). \"Pending\" = scheduled human task. \"Scheduled\" = scheduled AI task or AI task paused mid-execution. \"In-progress\" = currently running. \"Awaiting Trigger\" = waiting for a trigger event to fire. \"Input Needed\" = paused waiting for the user to provide input. \"Pending Approval\" = paused waiting for the user to approve an action. \"Completed\" = succeeded. \"Failed\" = errored or exited without completing. \"Canceled\" = explicitly canceled."
      },
      "contact_id": {
        "type": "number",
        "description": "Filter by associated contact"
      },
      "assignee_type": {
        "anyOf": [
          {
            "enum": [
              "ai",
              "user"
            ],
            "type": "string"
          },
          {
            "type": "null"
          }
        ],
        "description": "Filter by assignee type. Pass null to search both AI and human tasks — do that unless the user asked for one kind specifically."
      },
      "created_after": {
        "type": "string",
        "description": "ISO 8601 datetime — only return tasks created at or after this time."
      },
      "created_before": {
        "type": "string",
        "description": "ISO 8601 datetime — only return tasks created at or before this time."
      },
      "completed_after": {
        "type": "string",
        "description": "ISO 8601 datetime — only return tasks completed at or after this time."
      },
      "scheduled_after": {
        "type": "string",
        "description": "ISO 8601 datetime - only return tasks scheduled after this time. WARNING: excludes undated tasks (no scheduled date), which are common. Omit unless the user asked about a specific time window."
      },
      "assignee_user_id": {
        "type": "string",
        "description": "Filter by the user the task is assigned to (UUID of the user profile)."
      },
      "completed_before": {
        "type": "string",
        "description": "ISO 8601 datetime — only return tasks completed at or before this time."
      },
      "custom_object_id": {
        "type": "number",
        "description": "Filter by associated custom object"
      },
      "scheduled_before": {
        "type": "string",
        "description": "ISO 8601 datetime - only return tasks scheduled before this time. WARNING: excludes undated tasks (no scheduled date), which are common. Omit unless the user asked about a specific time window."
      },
      "contact_company_id": {
        "type": "number",
        "description": "Filter by associated company. Matches tasks linked directly to the company AND tasks linked to a contact who works there."
      },
      "created_by_user_id": {
        "type": "string",
        "description": "Filter by the user who created the task (UUID of the user profile)."
      }
    },
    "additionalProperties": false
  },
  "annotations": {
    "title": "Search tasks",
    "readOnlyHint": true,
    "openWorldHint": false,
    "destructiveHint": false
  }
}