Context & Examples

withContext

Provides domain-specific context and background knowledge to the agent. Context gives the agent essential information about the domain it’s operating in, including facts, conventions, operational details, and any other background knowledge needed to respond appropriately. This is critical for specialized agents that need to understand specific business rules, industry practices, or organizational details. Context differs from identity and capabilities:

Identity: Who/what the agent is
Capabilities: What the agent can do
Context: What the agent needs to know

withContext(text: string): this

text

string

required

Contextual information for the agent. Can be multi-line. Should contain factual, relevant information that helps the agent understand its operating environment.

Returns

this

The builder instance for method chaining

Examples

Medical Scheduling
E-commerce
Financial Services

builder.withContext(`
  Our clinic operates Monday-Friday, 9 AM to 5 PM.
  We have three doctors:
  - Dr. Smith specializes in general medicine
  - Dr. Jones specializes in cardiology
  - Dr. Lee specializes in pediatrics
  Average appointment duration is 30 minutes.
  Emergency appointments can be accommodated with 24-hour notice.
`);

builder.withContext(
  "We offer free shipping on orders over $50. " +
  "Standard delivery takes 3-5 business days. " +
  "Returns are accepted within 30 days of purchase."
);

builder.withContext(`
  Company policy:
  - All transactions require two-factor authentication
  - Daily withdrawal limit is $5,000
  - International transfers take 2-3 business days
  - Customer service available 24/7
`);

withExamples

Provides examples of desired agent behavior through few-shot learning. Examples are one of the most effective techniques for teaching an AI agent how to behave. By showing concrete input-output pairs, you demonstrate the exact pattern of responses you want. This is especially powerful for:

Establishing consistent response style
Demonstrating proper tool usage
Showing how to handle edge cases
Teaching domain-specific interaction patterns

Each example should demonstrate a specific aspect of good behavior. The model will learn from these examples and apply similar patterns in its responses.

withExamples(examples: Example[]): this

examples

Example[]

required

An array of Example objects showing input-output pairs. Use either user/assistant (conversational) or input/output (functional). Include explanation to clarify what the example demonstrates.

Returns

this

The builder instance for method chaining

Examples

Tool Usage
Response Style
Accumulating Examples

builder.withExamples([
  {
    user: "What's the weather in Paris?",
    assistant: "I'll check the weather for you. *calls get_weather tool with location: Paris*",
    explanation: "Shows proper tool invocation for weather queries"
  },
  {
    user: "Book me a table for 2 at 7pm",
    assistant: "I'll help you make a reservation. What restaurant would you like?",
    explanation: "Shows how to ask for missing required information"
  }
]);

builder.withExamples([
  {
    input: "Error: connection timeout",
    output: "I'm experiencing a connection issue. Let me try again in a moment.",
    explanation: "Demonstrates friendly error handling"
  }
]);

// Multiple calls accumulate examples
builder
  .withExamples([{ 
    user: "Hello", 
    assistant: "Hi! How can I help you today?" 
  }])
  .withExamples([{ 
    user: "Thanks", 
    assistant: "You're welcome! Let me know if you need anything else." 
  }]);

withErrorHandling

Defines how the agent should handle uncertainty, errors, and ambiguous situations. This method specifies the agent’s behavior when it encounters situations it’s not confident about, including:

Ambiguous user requests
Missing required information
Uncertainty about the correct response
Error conditions
Requests outside its capabilities

Clear error handling instructions help prevent:

Hallucinations (making up information when uncertain)
Inappropriate assumptions
Unhelpful or incorrect responses
Poor user experience during edge cases

withErrorHandling(instructions: string): this

instructions

string

required

Clear guidelines for handling uncertainty and errors. Should specify what the agent should do when it encounters various types of problematic situations. Can be multi-line for complex instructions.

Returns

this

The builder instance for method chaining

Examples

Basic
Comprehensive
Domain-Specific

builder.withErrorHandling(
  "When uncertain about a response, acknowledge your uncertainty and ask " +
  "clarifying questions rather than guessing or making assumptions."
);

builder.withErrorHandling(`
  Error Handling Guidelines:
  - If a request is ambiguous, ask specific clarifying questions
  - If you lack required information, explicitly list what's needed
  - If a request is outside your capabilities, clearly state your limitations
  - If uncertain about facts, acknowledge uncertainty rather than guessing
  - For tool errors, explain the issue in user-friendly terms and suggest alternatives
`);

builder.withErrorHandling(
  "If appointment slots are unavailable, suggest 2-3 alternative times nearby. " +
  "If a doctor is not available, suggest alternative doctors with similar specialties. " +
  "Always explain why a request cannot be fulfilled."
);

Core API

Builder Methods

Types

withContext

Examples

withExamples

Examples

withErrorHandling

Examples

Build docs developers (and LLMs) love

Core API

Builder Methods

Types

​withContext

​Examples

​withExamples

​Examples

​withErrorHandling

​Examples

Build docs developers (and LLMs) love

withContext

Examples

withExamples

Examples

withErrorHandling

Examples