Latitude MCP Server

/** * Phase 2 - Tier 2 Documentation Topics * * 13 new documentation topics: * - config-basics, config-advanced * - messages-roles, messages-multimodal * - tools-custom, tools-schema * - technique-few-shot, technique-cot, technique-tot * - recipe-generation, recipe-chatbot * - guide-safety, guide-performance */ export const DOCS_CONFIG_BASICS = `# Configuration Basics > Provider, model, and essential settings ## Overview Every prompt starts with a config section in YAML format. \\\`\\\`\\\`yaml --- provider: OpenAI model: gpt-4o --- Your prompt here... \\\`\\\`\\\` ## Required Settings | Setting | Description | Examples | |---------|-------------|----------| | \\\`provider\\\` | AI provider | OpenAI, Anthropic, Google | | \\\`model\\\` | Model ID | gpt-4o, claude-3-opus, gemini-pro | ## Common Providers | Provider | Popular Models | |----------|---------------| | OpenAI | gpt-4o, gpt-4o-mini, gpt-3.5-turbo | | Anthropic | claude-3-opus, claude-3-sonnet | | Google | gemini-pro, gemini-ultra | ## Next Steps - \\\`latitude_get_docs({ topic: "config-generation" })\\\` - \\\`latitude_get_docs({ topic: "structure" })\\\` `; export const DOCS_CONFIG_ADVANCED = `# Advanced Configuration > maxSteps, retries, headers, and more ## Advanced Settings | Setting | Description | Default | |---------|-------------|---------| | \\\`maxSteps\\\` | Max chain/agent steps | 20 | | \\\`maxRetries\\\` | Retry failed calls | 2 | | \\\`seed\\\` | Reproducible outputs | - | | \\\`stopSequences\\\` | Stop at strings | - | | \\\`headers\\\` | Custom HTTP headers | - | ## maxSteps Limit steps for chains and agents: \\\`\\\`\\\`yaml --- maxSteps: 50 --- \\\`\\\`\\\` ## stopSequences Stop generation at specific strings: \\\`\\\`\\\`yaml --- stopSequences: - "END" - "---" --- \\\`\\\`\\\` ## seed For reproducible outputs: \\\`\\\`\\\`yaml --- seed: 42 temperature: 0 --- \\\`\\\`\\\` ## Next Steps - \\\`latitude_get_docs({ topic: "chains" })\\\` - \\\`latitude_get_docs({ topic: "agents" })\\\` `; export const DOCS_MESSAGES_ROLES = `# Message Roles > System, user, assistant, tool messages ## Overview Messages define the conversation. Each has a role. ## System Message Text without tags is system message: \\\`\\\`\\\` You are a helpful assistant. \\\`\\\`\\\` Or explicit: \\\`\\\`\\\` <system> You are a helpful assistant. </system> \\\`\\\`\\\` ## User Message \\\`\\\`\\\` <user> What's the weather? </user> \\\`\\\`\\\` ## Assistant Message \\\`\\\`\\\` <assistant> I'd be happy to help! </assistant> \\\`\\\`\\\` ## Tool Message \\\`\\\`\\\` <tool id="call_123" name="get_weather"> {"temperature": 72} </tool> \\\`\\\`\\\` ## Message Order Messages form a conversation flow: 1. System (sets behavior) 2. User/Assistant (conversation) 3. Tool (function results) ## Next Steps - \\\`latitude_get_docs({ topic: "structure" })\\\` - \\\`latitude_get_docs({ topic: "conversation-history" })\\\` `; export const DOCS_MESSAGES_MULTIMODAL = `# Multimodal Messages > Images, files, and rich content ## Image Messages \\\`\\\`\\\` <user> <image url="{{ image_url }}" /> Describe this image. </user> \\\`\\\`\\\` ## Base64 Images \\\`\\\`\\\` <user> <image base64="{{ base64_data }}" type="image/png" /> What do you see? </user> \\\`\\\`\\\` ## File Attachments \\\`\\\`\\\` <user> <file url="{{ file_url }}" name="document.pdf" /> Summarize this document. </user> \\\`\\\`\\\` ## Model Support | Model | Images | Files | |-------|--------|-------| | gpt-4o | Yes | Via URL | | claude-3 | Yes | Yes | | gemini-pro | Yes | Yes | ## Tips - Use appropriate models (vision-capable) - Consider file size limits - Test with various formats ## Next Steps - \\\`latitude_get_docs({ topic: "messages-roles" })\\\` - \\\`latitude_get_docs({ topic: "config-basics" })\\\` `; export const DOCS_TOOLS_CUSTOM = `# Custom Tool Definition > Define your own tools with JSON Schema ## Overview Custom tools let the AI call your functions. \\\`\\\`\\\`yaml --- tools: - name: get_weather description: Get weather for a location parameters: type: object properties: location: type: string description: City name unit: type: string enum: [celsius, fahrenheit] required: [location] --- \\\`\\\`\\\` ## Tool Structure | Field | Description | |-------|-------------| | \\\`name\\\` | Function name (snake_case) | | \\\`description\\\` | What the tool does | | \\\`parameters\\\` | JSON Schema for inputs | ## Example: Database Query \\\`\\\`\\\`yaml tools: - name: query_database description: Run SQL query parameters: type: object properties: query: type: string limit: type: integer default: 10 required: [query] \\\`\\\`\\\` ## Tips - Clear descriptions help the model - Mark required parameters - Use enums for limited options ## Next Steps - \\\`latitude_get_docs({ topic: "tools-schema" })\\\` - \\\`latitude_get_docs({ topic: "tools-builtin" })\\\` `; export const DOCS_TOOLS_SCHEMA = `# Tool Schema Deep Dive > JSON Schema for parameters ## Basic Types | Type | Example | |------|---------| | \\\`string\\\` | "hello" | | \\\`number\\\` | 3.14 | | \\\`integer\\\` | 42 | | \\\`boolean\\\` | true | | \\\`array\\\` | [1, 2, 3] | | \\\`object\\\` | {...} | ## String Constraints \\\`\\\`\\\`yaml name: type: string minLength: 1 maxLength: 100 pattern: "^[A-Z].*" \\\`\\\`\\\` ## Number Constraints \\\`\\\`\\\`yaml age: type: integer minimum: 0 maximum: 150 \\\`\\\`\\\` ## Enum (Limited Options) \\\`\\\`\\\`yaml status: type: string enum: [pending, approved, rejected] \\\`\\\`\\\` ## Nested Objects \\\`\\\`\\\`yaml parameters: type: object properties: user: type: object properties: name: type: string email: type: string required: [name] \\\`\\\`\\\` ## Arrays \\\`\\\`\\\`yaml tags: type: array items: type: string minItems: 1 maxItems: 5 \\\`\\\`\\\` ## Next Steps - \\\`latitude_get_docs({ topic: "tools-custom" })\\\` - \\\`latitude_get_docs({ topic: "config-json-output" })\\\` `; export const DOCS_TECHNIQUE_FEW_SHOT = `# Few-Shot Learning > Examples for in-context learning ## Overview Provide examples to guide model behavior. ## Static Examples \\\`\\\`\\\` Translate to French: English: Hello French: Bonjour English: Goodbye French: Au revoir English: {{ input }} French: \\\`\\\`\\\` ## Dynamic Examples \\\`\\\`\\\` {{ for ex in examples }} Input: {{ ex.input }} Output: {{ ex.output }} {{ endfor }} Input: {{ new_input }} Output: \\\`\\\`\\\` ## Message-Based Examples \\\`\\\`\\\` {{ for ex in examples }} <user>{{ ex.question }}</user> <assistant>{{ ex.answer }}</assistant> {{ endfor }} <user>{{ user_question }}</user> \\\`\\\`\\\` ## Best Practices | Practice | Why | |----------|-----| | 3-5 examples | Enough to learn pattern | | Diverse examples | Cover edge cases | | Consistent format | Clear pattern | | Quality examples | Garbage in, garbage out | ## Next Steps - \\\`latitude_get_docs({ topic: "loops" })\\\` - \\\`latitude_get_docs({ topic: "recipe-classification" })\\\` `; export const DOCS_TECHNIQUE_COT = `# Chain-of-Thought > Step-by-step reasoning ## Overview Guide the model through reasoning steps. ## Zero-Shot CoT Just add the magic phrase: \\\`\\\`\\\` {{ problem }} Let's think step by step. \\\`\\\`\\\` ## Prompted CoT \\\`\\\`\\\` Solve this problem: {{ problem }} Let's work through this: 1. First, identify what we know 2. Then, consider the approach 3. Finally, calculate the answer \\\`\\\`\\\` ## CoT with Steps \\\`\\\`\\\` <step> # Understand the Problem {{ problem }} What is being asked? </step> <step> # Work Through Solution Based on my understanding, let me solve... </step> <step> # Verify and Conclude Let me check my work... </step> \\\`\\\`\\\` ## When to Use - Math problems - Logical reasoning - Complex analysis - Multi-step tasks ## Next Steps - \\\`latitude_get_docs({ topic: "technique-tot" })\\\` - \\\`latitude_get_docs({ topic: "chains" })\\\` `; export const DOCS_TECHNIQUE_TOT = `# Tree-of-Thoughts > Explore multiple reasoning paths ## Overview Generate multiple approaches, evaluate, select best. ## Basic ToT \\\`\\\`\\\` Problem: {{ problem }} ## Approach A 1. Reasoning... Confidence: X/10 ## Approach B 1. Reasoning... Confidence: X/10 ## Approach C 1. Reasoning... Confidence: X/10 ## Evaluation Compare approaches... ## Best Solution Based on Approach [X]: ... \\\`\\\`\\\` ## ToT with Chains \\\`\\\`\\\` <step> # Generate Approaches Problem: {{ problem }} Generate 3 different solutions. </step> <step> # Evaluate Each Rate feasibility, correctness, elegance (1-10). </step> <step> # Select and Develop Choose best and elaborate. </step> \\\`\\\`\\\` ## Multi-Agent ToT \\\`\\\`\\\`yaml type: agent agents: - agents/approach-a - agents/approach-b - agents/evaluator \\\`\\\`\\\` ## When to Use - Complex decisions - Creative problems - Planning tasks - When one path might fail ## Next Steps - \\\`latitude_get_docs({ topic: "technique-cot" })\\\` - \\\`latitude_get_docs({ topic: "agents" })\\\` `; export const DOCS_RECIPE_GENERATION = `# Content Generation > Writing and creation patterns ## Basic Generation \\\`\\\`\\\` Write a blog post about {{ topic }}. Requirements: - 500 words - Engaging tone - Include examples \\\`\\\`\\\` ## Structured Generation \\\`\\\`\\\`yaml schema: type: object properties: title: type: string introduction: type: string sections: type: array items: type: object properties: heading: type: string content: type: string conclusion: type: string \\\`\\\`\\\` ## Pipeline Pattern \\\`\\\`\\\` <step> # Research Topic: {{ topic }} Outline key points. </step> <step> # Draft Write first draft based on outline. </step> <step> # Edit Polish for clarity and engagement. </step> \\\`\\\`\\\` ## Tips - Set clear requirements - Use role prompting for voice - Structure with JSON schema - Pipeline for quality ## Next Steps - \\\`latitude_get_docs({ topic: "technique-role" })\\\` - \\\`latitude_get_docs({ topic: "chains" })\\\` `; export const DOCS_RECIPE_CHATBOT = `# Chatbot Patterns > Conversational AI design ## Basic Chatbot \\\`\\\`\\\` You are a helpful assistant for {{ company }}. Be friendly, concise, and helpful. {{ for msg in history }} <{{ msg.role }}>{{ msg.content }}</{{ msg.role }}> {{ endfor }} <user>{{ input }}</user> \\\`\\\`\\\` ## Persona Chatbot \\\`\\\`\\\` You are {{ persona.name }}, a {{ persona.role }}. Personality: {{ persona.traits | join(", ") }} Communication style: {{ persona.style }} Remember: {{ persona.memory }} {{ for msg in history }} <{{ msg.role }}>{{ msg.content }}</{{ msg.role }}> {{ endfor }} <user>{{ input }}</user> \\\`\\\`\\\` ## Task-Oriented Bot \\\`\\\`\\\` You are a customer service bot for {{ company }}. Available actions: - Check order status - Process returns - Answer FAQs Current context: - User: {{ user.name }} - Order: {{ user.recent_order }} {{ for msg in history }} <{{ msg.role }}>{{ msg.content }}</{{ msg.role }}> {{ endfor }} <user>{{ input }}</user> \\\`\\\`\\\` ## Tips - Set clear persona - Manage context length - Handle edge cases - Use memory wisely ## Next Steps - \\\`latitude_get_docs({ topic: "conversation-history" })\\\` - \\\`latitude_get_docs({ topic: "technique-role" })\\\` `; export const DOCS_GUIDE_SAFETY = `# Safety & Guardrails > Content filtering and prevention ## System Instructions \\\`\\\`\\\` You are a helpful assistant. IMPORTANT SAFETY RULES: - Never provide harmful information - Decline inappropriate requests politely - Stay within your knowledge boundaries - Recommend professionals for serious issues \\\`\\\`\\\` ## Content Boundaries \\\`\\\`\\\` You are an expert in {{ domain }}. You MUST NOT: - Provide medical/legal/financial advice - Share personal information - Generate harmful content - Pretend to be a human If asked about these, politely redirect. \\\`\\\`\\\` ## Input Validation \\\`\\\`\\\` {{ if input_seems_harmful }} I'm not able to help with that request. {{ else }} [Normal response] {{ endif }} \\\`\\\`\\\` ## Output Filtering \\\`\\\`\\\`yaml schema: type: object properties: response: type: string is_appropriate: type: boolean reasoning: type: string \\\`\\\`\\\` ## Best Practices | Practice | Description | |----------|-------------| | Clear boundaries | State what you won't do | | Polite refusals | Decline respectfully | | Redirect safely | Point to professionals | | Log concerns | Track problematic inputs | ## Next Steps - \\\`latitude_get_docs({ topic: "technique-role" })\\\` - \\\`latitude_get_docs({ topic: "guide-debugging" })\\\` `; export const DOCS_GUIDE_PERFORMANCE = `# Performance Optimization > Tokens, caching, efficiency ## Token Optimization ### Reduce Input - Shorter system prompts - Summarize context - Limit history length ### Reduce Output \\\`\\\`\\\`yaml maxTokens: 500 \\\`\\\`\\\` Or in prompt: \\\`\\\`\\\` Be concise. Maximum 2 sentences. \\\`\\\`\\\` ## Caching Enable caching with temperature 0: \\\`\\\`\\\`yaml temperature: 0 \\\`\\\`\\\` Identical inputs = cached responses. ## Model Selection | Need | Model | Tokens | |------|-------|--------| | Simple | gpt-4o-mini | Fewer | | Complex | gpt-4o | More | | Fast | gpt-3.5-turbo | Fewer | ## Batch Efficiently Process multiple items in one call: \\\`\\\`\\\` Analyze these 5 items: {{ for item in items }} {{ loop.index }}. {{ item }} {{ endfor }} Return analysis for each. \\\`\\\`\\\` ## Cost Estimation | Model | Input $/1M | Output $/1M | |-------|-----------|------------| | gpt-4o-mini | $0.15 | $0.60 | | gpt-4o | $2.50 | $10.00 | ## Tips - Start with smaller models - Use caching when possible - Batch similar requests - Monitor token usage ## Next Steps - \\\`latitude_get_docs({ topic: "config-generation" })\\\` - \\\`latitude_get_docs({ topic: "config-advanced" })\\\` `;