Tool tree
An assistant-building studio where a human and their agent work on the same surface, with every sensitive action behind a deterministic consent gate enforced outside the model.
Implementation
gradeNot graded
gradeNot graded
list_assistantsAnswer
List the assistants in the studio with the publish state of each one.
View tool JSON
{
"name": "list_assistants",
"kind": "answer",
"impl": "imperative",
"description": "List the assistants in the studio with the publish state of each one.",
"inputSchema": {
"type": "object",
"properties": {}
}
}create_assistantAction
Create a new assistant. It stays an unpublished draft until the user publishes it explicitly.
View tool JSON
{
"name": "create_assistant",
"kind": "act",
"impl": "imperative",
"description": "Create a new assistant. It stays an unpublished draft until the user publishes it explicitly.",
"inputSchema": {
"type": "object",
"properties": {
"name": {
"type": "string",
"description": "The assistant name as the customer sees it"
},
"purpose": {
"type": "string",
"description": "What it does, in one sentence"
},
"tone": {
"type": "string",
"enum": [
"formal",
"friendly",
"brief"
],
"description": "How the assistant speaks. Defaults to friendly."
},
"language": {
"type": "string",
"enum": [
"ar",
"en",
"tr"
],
"description": "The language the assistant replies in. Defaults to ar."
}
},
"required": [
"name",
"purpose"
]
}
}add_knowledgeAction
Append a snippet of text to an existing assistant's knowledge base.
View tool JSON
{
"name": "add_knowledge",
"kind": "act",
"impl": "imperative",
"description": "Append a snippet of text to an existing assistant's knowledge base.",
"inputSchema": {
"type": "object",
"properties": {
"id": {
"type": "string",
"description": "The assistant id, from list_assistants"
},
"text": {
"type": "string",
"description": "The text the assistant should know"
}
},
"required": [
"id",
"text"
]
}
}test_assistantAction
Send one test message to an assistant and get its reply. Changes nothing.
View tool JSON
{
"name": "test_assistant",
"kind": "act",
"impl": "imperative",
"description": "Send one test message to an assistant and get its reply. Changes nothing.",
"inputSchema": {
"type": "object",
"properties": {
"id": {
"type": "string",
"description": "The assistant id, from list_assistants"
},
"message": {
"type": "string",
"description": "The message to send, as a customer would"
}
},
"required": [
"id",
"message"
]
}
}