Skip to main content
Glama

Vibe Coder MCP

by freshtechbro
prompt-optimization.yaml•6.05 kB
# Prompt Optimization Configuration # Controls how LLM prompts are optimized for better JSON generation # Global optimization settings optimization: enabled: true max_prompt_length: 4000 # JSON-specific optimizations json_optimization: enabled: true include_schema_hints: true use_error_pattern_learning: true # Task-specific optimization rules task_patterns: - pattern: "module_selection" optimizations: - "json-system-enhancement" - "schema-hints" - "error-prevention-rules" - "large-number-handling" - pattern: "yaml_generation" optimizations: - "json-system-enhancement" - "schema-hints" - "error-prevention-rules" - "bracket-completion-hints" - pattern: "template_generation" optimizations: - "json-system-enhancement" - "error-prevention-rules" - "control-character-handling" - pattern: "intent_recognition" optimizations: - "json-system-enhancement" - "schema-hints" - "confidence-scoring-hints" # Error pattern learning configuration error_learning: enabled: true min_frequency_threshold: 2 # Only learn from errors that occur at least this many times max_patterns: 50 # Maximum number of error patterns to track # Known error patterns and their prevention rules known_patterns: missing_comma: frequency: 0 prevention_rule: "Always include commas between object properties" examples: - 'Use: {"key1": "value1", "key2": "value2"}' - 'Not: {"key1": "value1" "key2": "value2"}' control_character: frequency: 0 prevention_rule: "Escape control characters in strings (\\n, \\t, etc.)" examples: - 'Use: {"content": "line1\\nline2"}' - 'Not: {"content": "line1\nline2"}' trailing_comma: frequency: 0 prevention_rule: "Never include trailing commas after the last property" examples: - 'Use: {"key": "value"}' - 'Not: {"key": "value",}' single_quotes: frequency: 0 prevention_rule: "Always use double quotes, never single quotes" examples: - 'Use: {"key": "value"}' - 'Not: {\'key\': \'value\'}' large_numbers: frequency: 0 prevention_rule: "Use strings for numbers with more than 15 digits" examples: - 'Use: {"id": "12345678901234567890"}' - 'Not: {"id": 12345678901234567890}' # Prompt templates for different optimization types templates: json_system_enhancement: | CRITICAL JSON OUTPUT REQUIREMENTS: - You MUST respond with valid, parseable JSON only - Do NOT include markdown code blocks, backticks, or any formatting - Do NOT add explanatory text before or after the JSON - Start your response with { and end with } - Ensure all strings are properly quoted with double quotes - Ensure all object keys are strings in double quotes - Do NOT use trailing commas - Do NOT use single quotes - Escape special characters properly (\\n, \\t, \\", \\\\) json_user_enhancement: | OUTPUT FORMAT: Respond with a single, valid JSON object. No additional text, formatting, or explanations. output_format_instructions: | Your response must be a single JSON object that can be parsed by JSON.parse() without any modifications. schema_hints_template: | EXPECTED JSON STRUCTURE EXAMPLE: {schema} Ensure your response matches this structure exactly. error_prevention_rules: | JSON GENERATION RULES: - Never use single quotes - always use double quotes for strings and keys - Never include trailing commas after the last property - Always escape special characters in strings (\\n, \\t, \\", \\\\) - Never include comments in JSON output - Ensure all brackets and braces are properly matched - Use null instead of undefined for missing values - For large numbers (>15 digits), use strings to prevent precision loss - Ensure nested objects have proper comma separation - Validate that all required schema fields are included # Task-specific schema hints schemas: module_selection: type: "object" required: ["globalParams", "moduleSelections"] properties: globalParams: type: "object" properties: projectName: { type: "string" } projectDescription: { type: "string" } moduleSelections: type: "array" items: type: "object" properties: modulePath: { type: "string" } params: { type: "object" } moduleKey: { type: "string" } yaml_module: type: "object" required: ["moduleName", "description", "type"] properties: moduleName: { type: "string" } description: { type: "string" } type: { type: "string" } placeholders: type: "array" items: { type: "string" } provides: type: "object" properties: techStack: { type: "object" } directoryStructure: { type: "array" } dependencies: { type: "object" } setupCommands: { type: "array" } intent_recognition: type: "object" required: ["intent", "confidence", "parameters"] properties: intent: { type: "string" } confidence: { type: "number", minimum: 0, maximum: 1 } parameters: { type: "object" } context: { type: "object" } alternatives: { type: "array" } clarifications_needed: { type: "array" } # Performance monitoring monitoring: enabled: true log_optimization_stats: true track_success_rates: true alert_on_low_success_rate: true success_rate_threshold: 0.8 # Alert if success rate drops below 80% # Feature flags for gradual rollout feature_flags: enable_for_new_tasks: true enable_for_existing_tasks: true enable_schema_hints: true enable_error_learning: true enable_performance_monitoring: true

MCP directory API

We provide all the information about MCP servers via our MCP API.

curl -X GET 'https://glama.ai/api/mcp/v1/servers/freshtechbro/vibe-coder-mcp'

If you have feedback or need assistance with the MCP directory API, please join our Discord server