Prompts Reference

Prompts are guided workflow templates that MCP clients can surface as quick-start actions. When invoked, a prompt returns structured conversation context — a user turn and an assistant guidance turn — that initializes the conversation with the right intent and direction.

Prompts vs. Tools: Tools execute actions (create an agent, fetch data). Prompts return conversation context that guides the user through a workflow. You use a prompt to start a task; you use tools to complete it.

How MCP clients surface prompts: In Claude Desktop, Cursor, and other MCP-compatible clients, prompts appear as selectable quick-start actions. Selecting a prompt populates the conversation with the appropriate context, so the assistant knows exactly what to help with.

Gnosari provides 9 prompts in four categories:

Category	Count	Purpose
Replacement	4	Replace static forms with AI conversations
Utility	3	Generate agent configuration and content
Migration	1	Migrate from existing form tools
Embed	1	Configure chat widget embed for a website

---

Replacement Prompts

These prompts help you replace traditional static forms with AI-powered conversations that capture the same data — and often more.

---

replace_contact_form

Replace your contact form with an AI conversation that captures leads.

Tag: replacement

Parameters

Name	Type	Required	Description
business_name	str	Required	Name of the business whose contact form is being replaced
website_url	str	Optional	URL of the current website (default: "")
data_fields	str	Optional	Comma-separated list of data fields to collect (default: "")

Returns

list[Message] — 2 messages:

• User turn: Describes the business context and goal
• Assistant guidance turn: Provides step-by-step guidance for creating a contact agent with appropriate data collection configuration

Example

get_prompt("replace_contact_form", {
    "business_name": "Acme Corp",
    "website_url": "https://acme.com",
    "data_fields": "name, email, company, message"
})

---

replace_survey

Replace surveys with AI conversations that get better insights.

Tag: replacement

Parameters

Name	Type	Required	Description
survey_goal	str	Required	What the survey is trying to learn or measure
audience	str	Required	Who the survey is intended for

Returns

list[Message] — 2 messages:

• User turn: Describes the survey goal and target audience
• Assistant guidance turn: Provides step-by-step guidance for creating a conversational survey agent

Example

get_prompt("replace_survey", {
    "survey_goal": "Understand customer satisfaction with our onboarding",
    "audience": "New customers who signed up in the last 30 days"
})

---

replace_booking_form

Replace your booking form with AI-powered scheduling conversations.

Tag: replacement

Parameters

Name	Type	Required	Description
service_type	str	Required	Type of service being booked (e.g., "consulting call", "haircut", "medical appointment")
business_name	str	Required	Name of the business offering the service

Returns

list[Message] — 2 messages:

• User turn: Describes the booking context and service type
• Assistant guidance turn: Provides step-by-step guidance for creating a booking agent with appropriate fields

Example

get_prompt("replace_booking_form", {
    "service_type": "30-minute strategy call",
    "business_name": "Growth Consulting"
})

---

replace_job_application

Replace job application forms with AI candidate screening.

Tag: replacement

Parameters

Name	Type	Required	Description
role_title	str	Required	Title of the job role being filled
company_name	str	Required	Name of the hiring company
skills	str	Optional	Comma-separated list of required or desired skills (default: "")

Returns

list[Message] — 2 messages:

• User turn: Describes the role and company context
• Assistant guidance turn: Provides step-by-step guidance for creating a candidate screening agent

Example

get_prompt("replace_job_application", {
    "role_title": "Senior Backend Engineer",
    "company_name": "Acme Corp",
    "skills": "Python, FastAPI, PostgreSQL"
})

---

Utility Prompts

These prompts help you generate configuration and content for your Gnosari agents.

---

write_agent_instructions

Generate system instructions for a Gnosari agent.

Tag: utility

Parameters

Name	Type	Required	Description
use_case	str	Required	What the agent will be used for
business_context	str	Required	Description of the business and its context
tone	str	Optional	Desired communication tone (e.g., "professional", "casual", "empathetic") (default: "")

Returns

list[Message] — 2 messages:

• User turn: Describes the use case, business context, and tone requirements
• Assistant guidance turn: Provides a structured approach and example instructions for the agent

Example

get_prompt("write_agent_instructions", {
    "use_case": "Qualify inbound sales leads",
    "business_context": "B2B SaaS company selling project management software to teams of 10-200",
    "tone": "professional but approachable"
})

---

design_data_collection

Design a data collection template for your agent.

Tag: utility

Parameters

Name	Type	Required	Description
what_to_collect	str	Required	Description of what data the agent should collect from conversations
collection_mode	str	Optional	How aggressively to collect data: passive, opportunistic, active, or guided (default: "")

Returns

list[Message] — 2 messages:

• User turn: Describes the data collection goal and preferences
• Assistant guidance turn: Provides a recommended template structure with field definitions and collection mode guidance

Example

get_prompt("design_data_collection", {
    "what_to_collect": "Contact information and budget range for B2B leads",
    "collection_mode": "opportunistic"
})

---

suggest_conversation_starters

Generate conversation starters for your agent.

Tag: utility

Parameters

Name	Type	Required	Description
agent_purpose	str	Required	What the agent does and who it serves
count	str	Optional	Number of starters to generate (default: "")

Note: count is a str type, not an integer.

Returns

list[Message] — 2 messages:

• User turn: Describes the agent purpose and the number of starters needed
• Assistant guidance turn: Provides a set of conversation starters appropriate for the agent's context

Example

get_prompt("suggest_conversation_starters", {
    "agent_purpose": "Help visitors explore pricing options for a SaaS product",
    "count": "5"
})

---

Migration Prompts

These prompts help you migrate from existing form and survey tools to Gnosari AI conversations.

---

migrate_from_typeform

Migrate from Typeform or Google Forms to a Gnosari AI conversation.

Tag: migration

This is the only prompt where all parameters are optional. You can invoke it with no arguments and the assistant will guide you through the migration process interactively.

Parameters

Name	Type	Required	Description
form_url	str	Optional	URL of the existing Typeform or Google Form (default: "")
field_count	str	Optional	Number of fields in the existing form (default: "")

Returns

list[Message] — 2 messages:

• User turn: Describes the migration goal with any provided form context
• Assistant guidance turn: Provides a step-by-step migration plan including data mapping, agent creation, and rollout strategy

Example — with context

get_prompt("migrate_from_typeform", {
    "form_url": "https://form.typeform.com/to/abc123",
    "field_count": "8"
})

Example — no arguments

get_prompt("migrate_from_typeform", {})

---

Embed Prompts

These prompts help you configure the Gnosari chat widget for embedding in a website.

---

embed_agent

Recommend the right embed configuration for a website based on its use case.

Tag: embed

Parameters

Name	Type	Required	Description
use_case	str	Required	Website use case (e.g., "support", "docs", "app", "mobile")
agent_name	str	Optional	Name of the agent to embed (default: "")

Returns

list[Message] — 2 messages:

• User turn: Describes the website use case (and agent name if provided)
• Assistant guidance turn: Recommends the appropriate display mode and configuration (support/sales→bubble, docs→sidebar, app→sidebar-push, mobile→drawer)

Example

get_prompt("embed_agent", {
    "use_case": "customer support on a SaaS dashboard",
    "agent_name": "Support Bot"
})

Example — without agent name

get_prompt("embed_agent", {
    "use_case": "documentation site"
})

---

Prompt Summary

Prompt	Tag	Required Params	Optional Params
replace_contact_form	replacement	business_name	website_url, data_fields
replace_survey	replacement	survey_goal, audience	—
replace_booking_form	replacement	service_type, business_name	—
replace_job_application	replacement	role_title, company_name	skills
write_agent_instructions	utility	use_case, business_context	tone
design_data_collection	utility	what_to_collect	collection_mode
suggest_conversation_starters	utility	agent_purpose	count
migrate_from_typeform	migration	—	form_url, field_count
embed_agent	embed	use_case	agent_name

All parameters are str type. Optional parameters default to "" (empty string).

---

Next Steps

• Agent Management Tools - Create and configure agents
• Guides - Common workflows using prompts and tools together