\n \n \n \n \n `);\n }\n});\n\n// Serve frontend files\napp.get(\"/frontend/*\", c => serveFile(c.req.path, import.meta.url));\n\n// Root endpoint with API info\napp.get(\"/\", (c) => {\n return c.json({\n name: \"Anthropic Agent API with MCPs\",\n version: \"1.0.0\",\n description: \"API completa para gerenciar agentes Anthropic com suporte a MCPs e workflows assíncronos\",\n endpoints: {\n health: \"/health\",\n swagger: \"/swagger\",\n api: {\n agents: \"/api/agents\",\n workflows: \"/api/workflows\",\n executions: \"/api/executions\",\n docs: \"/api/docs\"\n }\n },\n features: [\n \"🤖 Agentes Anthropic com Claude Sonnet 4\",\n \"🧠 Thinking mode ativo\",\n \"🔗 MCPs: Firecrawl, Serper.dev, Semrush\",\n \"🔄 Workflows com pipeline de agentes\",\n \"⚡ Execução assíncrona\",\n \"📊 Monitoramento e estatísticas\",\n \"📚 Documentação Swagger interativa\"\n ],\n mcps_supported: [\n {\n name: \"Firecrawl\",\n description: \"Web scraping e extração de conteúdo\",\n tools: [\"firecrawl_scrape\"]\n },\n {\n name: \"Serper.dev\",\n description: \"Pesquisa no Google (web, imagens, vídeos, etc.)\",\n tools: [\"serper_search\"]\n },\n {\n name: \"Semrush\",\n description: \"Análise de SEO e palavras-chave\",\n tools: [\"semrush_domain_analysis\"]\n }\n ]\n });\n});\n\n// 404 handler\napp.notFound((c) => {\n return c.json({\n success: false,\n error: \"Endpoint not found\",\n available_endpoints: [\n \"GET /\",\n \"GET /health\",\n \"GET /swagger\",\n \"GET /api/agents\",\n \"POST /api/agents\",\n \"GET /api/workflows\",\n \"POST /api/workflows\",\n \"GET /api/executions\"\n ]\n }, 404);\n});\n\n// Error handler\napp.onError((err, c) => {\n console.error(\"API Error:\", err);\n return c.json({\n success: false,\n error: err.message || \"Internal server error\",\n timestamp: new Date().toISOString()\n }, 500);\n});\n\nexport default app.fetch;",[],[150,274],[150],{"_102":285,"_95":286,"_151":287,"_153":-5,"_154":136,"_155":288,"_157":158,"_104":289,"_99":32,"_160":161,"_162":-5,"_163":22,"_164":-5,"_165":290,"_98":302},"d600c552-8dc8-11f0-8e44-0224a6c84d84","frontend","43e310b2-5d9c-4801-ba80-f02b96d25a25",13,["D",1757455417170],[291],{"_102":292,"_95":293,"_151":294,"_153":287,"_154":136,"_155":295,"_157":158,"_104":296,"_99":297,"_160":298,"_162":299,"_163":22,"_164":-5,"_165":300,"_98":301},"041626ce-8dcb-11f0-8a1a-0224a6c84d84","swagger.html","d600a086-8dc8-11f0-8e44-0224a6c84d84",49,["D",1757456353474],1,"file","\n\n\n \n \n Anthropic Agent API - Swagger UI\n \n \n\n\n

🤖 Anthropic Agent API

API completa para gerenciar agentes Anthropic com MCPs e workflows assíncronos

\n \n

🧠 Agentes Inteligentes

Claude Sonnet 4 com thinking mode
1M tokens de contexto
64k tokens de output máximo
32k budget tokens
System prompts personalizáveis

\n \n

🔗 MCPs Integrados

Firecrawl Web scraping
Serper.dev Google Search
Semrush SEO Analysis

\n \n

🔄 Workflows Avançados

Pipeline de agentes sequenciais
Mapeamento de input/output
Execução assíncrona
Monitoramento em tempo real

\n \n

📊 Monitoramento

Status de execuções em tempo real
Logs detalhados
Estatísticas de uso
Retry de execuções falhadas

\n \n

⚠️ Variáveis de Ambiente Necessárias

ANTHROPIC_API_KEY - Chave da API Anthropic
FIRECRAWL_API_KEY - Chave da API Firecrawl
SERPER_API_KEY - Chave da API Serper.dev
SEMRUSH_API_KEY - Chave da API Semrush

\n \n

🚀 Quick Start

1. Criar um Agente de Pesquisa Web:

\nPOST /api/agents\nContent-Type: application/json\n\n{\n  \"name\": \"Web Research Agent\",\n  \"description\": \"Agente especializado em pesquisa web\",\n  \"system_prompt\": \"Você é um especialista em pesquisa web. Use as ferramentas disponíveis (Firecrawl para scraping e Serper para busca) para encontrar informações relevantes e precisas. Sempre cite suas fontes e forneça análises detalhadas.\",\n  \"mcps\": [\n    {\n      \"type\": \"firecrawl\",\n      \"enabled\": true,\n      \"config\": {\n        \"formats\": [\"markdown\"],\n        \"onlyMainContent\": true\n      }\n    },\n    {\n      \"type\": \"serper\",\n      \"enabled\": true,\n      \"config\": {\n        \"type\": \"search\",\n        \"location\": \"Brazil\"\n      }\n    }\n  ]\n}\n

\n \n

2. Executar o Agente:

\nPOST /api/agents/{agent_id}/execute\nContent-Type: application/json\n\n{\n  \"input\": \"Pesquise sobre as últimas tendências em inteligência artificial para 2024\",\n  \"context\": {\n    \"user_id\": \"user123\",\n    \"preferences\": {\n      \"detail_level\": \"high\",\n      \"language\": \"pt-BR\"\n    }\n  }\n}\n

\n \n

3. Verificar Status da Execução:

\n \nGET /api/executions/{execution_id}/status\n \n

\n \n

📊 Exemplos de Input por Tipo de Agente

\n \n

🔍 Agente de Pesquisa Web (Firecrawl + Serper)

\n{\n  \"input\": {\n    \"query\": \"inteligência artificial 2024\",\n    \"focus\": \"tendências e inovações\",\n    \"sources_needed\": 5,\n    \"language\": \"português\"\n  }\n}\n\n// Ou para scraping específico:\n{\n  \"input\": {\n    \"task\": \"scrape_and_analyze\",\n    \"url\": \"https://example.com/article\",\n    \"analysis_type\": \"content_summary\"\n  }\n}\n

\n \n

📈 Agente SEO (Semrush + Serper)

\n{\n  \"input\": {\n    \"domain\": \"example.com\",\n    \"analysis_type\": \"keyword_research\",\n    \"competitors\": [\"competitor1.com\", \"competitor2.com\"],\n    \"target_market\": \"Brazil\"\n  }\n}\n

\n \n

🔄 Workflow de Múltiplos Agentes

\n// 1. Criar o workflow:\nPOST /api/workflows\n{\n  \"name\": \"Research and Analysis Pipeline\",\n  \"description\": \"Pipeline de pesquisa e análise\",\n  \"agents\": [\n    {\n      \"agent_id\": \"research-agent-id\",\n      \"order\": 1,\n      \"input_mapping\": {},\n      \"output_mapping\": {\n        \"research_results\": \"text\",\n        \"sources\": \"sources\"\n      }\n    },\n    {\n      \"agent_id\": \"analysis-agent-id\", \n      \"order\": 2,\n      \"input_mapping\": {\n        \"data_to_analyze\": \"research_results\",\n        \"source_urls\": \"sources\"\n      },\n      \"output_mapping\": {}\n    }\n  ]\n}\n\n// 2. Executar o workflow:\nPOST /api/workflows/{workflow_id}/execute\n{\n  \"input\": {\n    \"topic\": \"inteligência artificial 2024\",\n    \"research_depth\": \"comprehensive\",\n    \"target_audience\": \"desenvolvedores\"\n  }\n}\n

\n \n

\n \n \n \n \n\n",[],[286,293],[286],{"_102":304,"_95":305,"_151":306,"_153":-5,"_154":136,"_155":307,"_157":158,"_104":308,"_99":32,"_160":161,"_162":-5,"_163":22,"_164":-5,"_165":309,"_98":319},"fe0f2f80-8dc7-11f0-ada3-0224a6c84d84","shared","9e4dad1d-e23d-43f5-b53a-9e3d233afdb8",2,["D",1757455054884],[310],{"_102":311,"_95":312,"_151":313,"_153":306,"_154":136,"_155":314,"_157":158,"_104":315,"_99":297,"_160":178,"_162":316,"_163":22,"_164":-5,"_165":317,"_98":318},"781827d8-8dcc-11f0-8aad-0224a6c84d84","types.ts","fe0f2576-8dc7-11f0-ada3-0224a6c84d84",57,["D",1757456977608],"// Tipos compartilhados entre frontend e backend\n\nexport interface Agent {\n id: string;\n name: string;\n description: string;\n model: string;\n max_tokens: number;\n context_window: number;\n thinking_enabled: boolean;\n budget_tokens: number;\n system_prompt: string;\n mcps: MCPConfig[];\n created_at: string;\n updated_at: string;\n}\n\nexport interface MCPConfig {\n type: 'firecrawl' | 'serper' | 'semrush';\n enabled: boolean;\n config: Record;\n}\n\nexport interface Workflow {\n id: string;\n name: string;\n description: string;\n agents: WorkflowAgent[];\n created_at: string;\n updated_at: string;\n}\n\nexport interface WorkflowAgent {\n agent_id: string;\n order: number;\n input_mapping?: Record;\n output_mapping?: Record;\n}\n\nexport interface Execution {\n id: string;\n type: 'agent' | 'workflow';\n target_id: string;\n status: 'pending' | 'running' | 'completed' | 'failed' | 'cancelled';\n input: any;\n output?: any;\n error?: string;\n progress?: number;\n started_at?: string;\n completed_at?: string;\n created_at: string;\n}\n\nexport interface AgentExecution {\n id: string;\n agent_id: string;\n execution_id: string;\n status: 'pending' | 'running' | 'completed' | 'failed';\n input: any;\n output?: any;\n error?: string;\n tokens_used?: number;\n thinking_content?: string;\n started_at?: string;\n completed_at?: string;\n created_at: string;\n}\n\nexport interface CreateAgentRequest {\n name: string;\n description: string;\n system_prompt: string;\n mcps: MCPConfig[];\n}\n\nexport interface CreateWorkflowRequest {\n name: string;\n description: string;\n agents: Omit & { agent_id: string }[];\n}\n\nexport interface ExecuteAgentRequest {\n input: any;\n}\n\nexport interface ExecuteWorkflowRequest {\n input: any;\n}\n\nexport interface MCPResponse {\n success: boolean;\n data?: any;\n error?: string;\n}\n\nexport interface FirecrawlConfig {\n formats?: string[];\n includeTags?: string[];\n excludeTags?: string[];\n onlyMainContent?: boolean;\n}\n\nexport interface SerperConfig {\n type?: 'search' | 'images' | 'videos' | 'places' | 'maps' | 'news' | 'shopping';\n location?: string;\n gl?: string;\n hl?: string;\n}\n\nexport interface SemrushConfig {\n database?: string;\n display_limit?: number;\n export_columns?: string;\n}",[],[305,312],[305],{"_102":321,"_95":322,"_151":323,"_153":-5,"_154":136,"_155":135,"_157":158,"_104":324,"_99":307,"_160":298,"_162":325,"_163":22,"_164":-5,"_165":326,"_98":327},"0a62a3b6-8dcd-11f0-87b2-0224a6c84d84","EXAMPLES.md","2aeff23e-8dcb-11f0-8ceb-0224a6c84d84",["D",1757457223044],"# 📚 Guia Completo de Exemplos - Anthropic Agent API\n\nEste guia contém exemplos práticos de como usar os bodies das requests para cada endpoint da API.\n\n## 🤖 Criando Agentes\n\n### 1. Agente de Pesquisa Web (Firecrawl + Serper)\n\n```json\nPOST /api/agents\nContent-Type: application/json\n\n{\n \"name\": \"Web Research Agent\",\n \"description\": \"Agente especializado em pesquisa web e análise de conteúdo\",\n \"system_prompt\": \"Você é um especialista em pesquisa web. Use as ferramentas disponíveis (Firecrawl para scraping e Serper para busca) para encontrar informações relevantes e precisas. Sempre cite suas fontes e forneça análises detalhadas.\",\n \"mcps\": [\n {\n \"type\": \"firecrawl\",\n \"enabled\": true,\n \"config\": {\n \"formats\": [\"markdown\"],\n \"onlyMainContent\": true,\n \"excludeTags\": [\"script\", \"style\", \"nav\"]\n }\n },\n {\n \"type\": \"serper\",\n \"enabled\": true,\n \"config\": {\n \"type\": \"search\",\n \"location\": \"Brazil\",\n \"gl\": \"br\",\n \"hl\": \"pt\"\n }\n }\n ]\n}\n```\n\n### 2. Agente SEO (Semrush + Serper)\n\n```json\nPOST /api/agents\nContent-Type: application/json\n\n{\n \"name\": \"SEO Analyst Agent\",\n \"description\": \"Especialista em análise de SEO e palavras-chave\",\n \"system_prompt\": \"Você é um especialista em SEO. Use o Semrush para analisar domínios, palavras-chave e competidores. Forneça insights acionáveis e recomendações estratégicas baseadas em dados.\",\n \"mcps\": [\n {\n \"type\": \"semrush\",\n \"enabled\": true,\n \"config\": {\n \"database\": \"br\",\n \"display_limit\": 20,\n \"export_columns\": \"Ph,Po,Pp,Pd,Nq,Cp,Ur,Tr,Tc,Co,Nr,Td\"\n }\n },\n {\n \"type\": \"serper\",\n \"enabled\": true,\n \"config\": {\n \"type\": \"search\",\n \"location\": \"Brazil\"\n }\n }\n ]\n}\n```\n\n### 3. Agente Completo (Todos os MCPs)\n\n```json\nPOST /api/agents\nContent-Type: application/json\n\n{\n \"name\": \"Complete Research Agent\",\n \"description\": \"Agente completo com todos os MCPs disponíveis\",\n \"system_prompt\": \"Você é um pesquisador completo com acesso a ferramentas de web scraping (Firecrawl), busca (Serper) e análise SEO (Semrush). Use essas ferramentas de forma inteligente para fornecer análises abrangentes e insights valiosos.\",\n \"mcps\": [\n {\n \"type\": \"firecrawl\",\n \"enabled\": true,\n \"config\": {\n \"formats\": [\"markdown\", \"html\"],\n \"includeTags\": [\"p\", \"h1\", \"h2\", \"h3\", \"h4\", \"ul\", \"ol\", \"li\"],\n \"onlyMainContent\": true\n }\n },\n {\n \"type\": \"serper\",\n \"enabled\": true,\n \"config\": {\n \"type\": \"search\",\n \"location\": \"Brazil\"\n }\n },\n {\n \"type\": \"semrush\",\n \"enabled\": true,\n \"config\": {\n \"database\": \"us\",\n \"display_limit\": 15\n }\n }\n ]\n}\n```\n\n## ⚡ Executando Agentes\n\n### 1. Input Simples (String)\n\n```json\nPOST /api/agents/{agent_id}/execute\nContent-Type: application/json\n\n{\n \"input\": \"Pesquise sobre as últimas tendências em inteligência artificial para 2024\"\n}\n```\n\n### 2. Input com Contexto Adicional\n\n```json\nPOST /api/agents/{agent_id}/execute\nContent-Type: application/json\n\n{\n \"input\": {\n \"content\": \"Pesquise sobre as últimas tendências em inteligência artificial para 2024\",\n \"context\": {\n \"user_id\": \"user123\",\n \"project_id\": \"proj456\",\n \"preferences\": {\n \"detail_level\": \"high\",\n \"include_sources\": true,\n \"format\": \"structured_report\"\n }\n }\n }\n}\n```\n\n### 3. Input Estruturado para Pesquisa Web\n\n```json\nPOST /api/agents/{agent_id}/execute\nContent-Type: application/json\n\n{\n \"input\": {\n \"query\": \"inteligência artificial 2024\",\n \"focus\": \"tendências e inovações\",\n \"sources_needed\": 5,\n \"language\": \"português\",\n \"depth\": \"comprehensive\",\n \"context\": {\n \"user_id\": \"user123\",\n \"project_id\": \"proj456\",\n \"preferences\": {\n \"detail_level\": \"high\",\n \"include_sources\": true,\n \"format\": \"structured_report\"\n }\n }\n }\n}\n```\n\n### 4. Input para Web Scraping\n\n```json\nPOST /api/agents/{agent_id}/execute\nContent-Type: application/json\n\n{\n \"input\": {\n \"task\": \"scrape_and_analyze\",\n \"url\": \"https://example.com/article-about-ai\",\n \"analysis_type\": \"content_summary\",\n \"extract_keywords\": true,\n \"sentiment_analysis\": true,\n \"context\": {\n \"purpose\": \"competitive_analysis\",\n \"target_audience\": \"developers\"\n }\n }\n}\n```\n\n### 5. Input para Análise SEO\n\n```json\nPOST /api/agents/{agent_id}/execute\nContent-Type: application/json\n\n{\n \"input\": {\n \"domain\": \"example.com\",\n \"analysis_type\": \"keyword_research\",\n \"competitors\": [\"competitor1.com\", \"competitor2.com\"],\n \"target_market\": \"Brazil\",\n \"focus_keywords\": [\"ia\", \"machine learning\", \"automação\"],\n \"context\": {\n \"campaign_goal\": \"content_strategy\",\n \"budget\": \"medium\",\n \"timeline\": \"Q1_2024\"\n }\n }\n}\n```\n\n### 6. Input com Mensagens (Conversação)\n\n```json\nPOST /api/agents/{agent_id}/execute\nContent-Type: application/json\n\n{\n \"input\": {\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": \"Preciso de uma análise completa sobre IA em 2024, focando em aplicações práticas para empresas brasileiras\"\n }\n ],\n \"context\": {\n \"session_id\": \"session789\",\n \"previous_research\": {\n \"topics\": [\"machine learning\", \"neural networks\"],\n \"date\": \"2024-01-15\"\n }\n }\n }\n}\n```\n\n## 🔄 Criando Workflows\n\n### 1. Workflow de Pesquisa e Análise\n\n```json\nPOST /api/workflows\nContent-Type: application/json\n\n{\n \"name\": \"Research and Analysis Pipeline\",\n \"description\": \"Pipeline que primeiro pesquisa informações na web e depois analiza os resultados encontrados\",\n \"agents\": [\n {\n \"agent_id\": \"research-agent-id\",\n \"order\": 1,\n \"input_mapping\": {},\n \"output_mapping\": {\n \"research_results\": \"text\",\n \"sources\": \"sources\",\n \"keywords\": \"extracted_keywords\"\n }\n },\n {\n \"agent_id\": \"analysis-agent-id\",\n \"order\": 2,\n \"input_mapping\": {\n \"data_to_analyze\": \"research_results\",\n \"source_urls\": \"sources\",\n \"key_terms\": \"keywords\"\n },\n \"output_mapping\": {\n \"final_analysis\": \"analysis\",\n \"recommendations\": \"recommendations\"\n }\n }\n ]\n}\n```\n\n### 2. Workflow SEO + Criação de Conteúdo\n\n```json\nPOST /api/workflows\nContent-Type: application/json\n\n{\n \"name\": \"SEO Analysis + Content Creation\",\n \"description\": \"Analisa palavras-chave de um domínio e cria conteúdo otimizado\",\n \"agents\": [\n {\n \"agent_id\": \"seo-agent-id\",\n \"order\": 1,\n \"input_mapping\": {},\n \"output_mapping\": {\n \"top_keywords\": \"keywords.high_volume\",\n \"competitor_analysis\": \"competitors.analysis\",\n \"content_gaps\": \"opportunities.content_gaps\"\n }\n },\n {\n \"agent_id\": \"content-agent-id\",\n \"order\": 2,\n \"input_mapping\": {\n \"target_keywords\": \"top_keywords\",\n \"competitor_insights\": \"competitor_analysis\",\n \"content_opportunities\": \"content_gaps\"\n },\n \"output_mapping\": {}\n }\n ]\n}\n```\n\n### 3. Workflow de Análise Competitiva Completa\n\n```json\nPOST /api/workflows\nContent-Type: application/json\n\n{\n \"name\": \"Complete Competitive Analysis\",\n \"description\": \"Análise completa de concorrentes incluindo SEO, conteúdo e estratégias\",\n \"agents\": [\n {\n \"agent_id\": \"web-research-agent-id\",\n \"order\": 1,\n \"input_mapping\": {},\n \"output_mapping\": {\n \"competitor_info\": \"research_results\",\n \"market_data\": \"market_analysis\"\n }\n },\n {\n \"agent_id\": \"seo-agent-id\",\n \"order\": 2,\n \"input_mapping\": {\n \"domains_to_analyze\": \"competitor_info.domains\"\n },\n \"output_mapping\": {\n \"seo_insights\": \"seo_analysis\",\n \"keyword_gaps\": \"keyword_opportunities\"\n }\n },\n {\n \"agent_id\": \"strategy-agent-id\",\n \"order\": 3,\n \"input_mapping\": {\n \"market_research\": \"market_data\",\n \"seo_data\": \"seo_insights\",\n \"opportunities\": \"keyword_gaps\"\n },\n \"output_mapping\": {}\n }\n ]\n}\n```\n\n## 🚀 Executando Workflows\n\n### 1. Workflow de Pesquisa\n\n```json\nPOST /api/workflows/{workflow_id}/execute\nContent-Type: application/json\n\n{\n \"input\": {\n \"topic\": \"inteligência artificial 2024\",\n \"research_depth\": \"comprehensive\",\n \"target_audience\": \"desenvolvedores\",\n \"output_format\": \"detailed_report\",\n \"language\": \"português\"\n },\n \"context\": {\n \"project_id\": \"proj123\",\n \"deadline\": \"2024-02-01\",\n \"budget\": \"high\",\n \"stakeholders\": [\"CTO\", \"Product Manager\"]\n }\n}\n```\n\n### 2. Workflow SEO\n\n```json\nPOST /api/workflows/{workflow_id}/execute\nContent-Type: application/json\n\n{\n \"input\": {\n \"domain\": \"example.com\",\n \"target_keywords\": [\"ia\", \"machine learning\", \"automação\"],\n \"content_type\": \"blog_post\",\n \"word_count\": 1500,\n \"target_market\": \"Brazil\"\n },\n \"context\": {\n \"campaign_name\": \"AI_Content_Q1\",\n \"brand_voice\": \"technical_but_accessible\",\n \"seo_goals\": [\"increase_organic_traffic\", \"improve_rankings\"]\n }\n}\n```\n\n### 3. Workflow de Análise Competitiva\n\n```json\nPOST /api/workflows/{workflow_id}/execute\nContent-Type: application/json\n\n{\n \"input\": {\n \"company\": \"MinhaEmpresa\",\n \"industry\": \"tecnologia\",\n \"competitors\": [\"competitor1.com\", \"competitor2.com\", \"competitor3.com\"],\n \"analysis_focus\": [\"pricing\", \"features\", \"marketing\", \"seo\"],\n \"market_region\": \"Brazil\"\n },\n \"context\": {\n \"analysis_purpose\": \"strategic_planning\",\n \"quarter\": \"Q1_2024\",\n \"team\": \"product_strategy\"\n }\n}\n```\n\n## 📊 Configurações Avançadas de MCPs\n\n### Firecrawl - Configurações Detalhadas\n\n```json\n{\n \"type\": \"firecrawl\",\n \"enabled\": true,\n \"config\": {\n \"formats\": [\"markdown\", \"html\"],\n \"includeTags\": [\"p\", \"h1\", \"h2\", \"h3\", \"h4\", \"h5\", \"h6\", \"ul\", \"ol\", \"li\", \"blockquote\", \"table\"],\n \"excludeTags\": [\"script\", \"style\", \"nav\", \"footer\", \"aside\", \"advertisement\"],\n \"onlyMainContent\": true\n }\n}\n```\n\n### Serper - Diferentes Tipos de Busca\n\n```json\n// Busca Web\n{\n \"type\": \"serper\",\n \"enabled\": true,\n \"config\": {\n \"type\": \"search\",\n \"location\": \"Brazil\",\n \"gl\": \"br\",\n \"hl\": \"pt\"\n }\n}\n\n// Busca de Imagens\n{\n \"type\": \"serper\",\n \"enabled\": true,\n \"config\": {\n \"type\": \"images\",\n \"location\": \"Brazil\"\n }\n}\n\n// Busca de Notícias\n{\n \"type\": \"serper\",\n \"enabled\": true,\n \"config\": {\n \"type\": \"news\",\n \"location\": \"Brazil\",\n \"hl\": \"pt\"\n }\n}\n```\n\n### Semrush - Configurações por Região\n\n```json\n// Brasil\n{\n \"type\": \"semrush\",\n \"enabled\": true,\n \"config\": {\n \"database\": \"br\",\n \"display_limit\": 20,\n \"export_columns\": \"Ph,Po,Pp,Pd,Nq,Cp,Ur,Tr,Tc,Co,Nr,Td\"\n }\n}\n\n// Estados Unidos\n{\n \"type\": \"semrush\",\n \"enabled\": true,\n \"config\": {\n \"database\": \"us\",\n \"display_limit\": 50,\n \"export_columns\": \"Ph,Po,Pp,Pd,Nq,Cp,Ur,Tr,Tc,Co,Nr,Td\"\n }\n}\n```\n\n## 🔍 Monitoramento de Execuções\n\n### Verificar Status\n\n```bash\nGET /api/executions/{execution_id}\nGET /api/executions/{execution_id}/status\nGET /api/executions/{execution_id}/logs\n```\n\n### Listar Execuções com Filtros\n\n```bash\nGET /api/executions?type=agent&status=running\nGET /api/executions?type=workflow&status=completed&limit=10\n```\n\n### Cancelar Execução\n\n```bash\nPOST /api/executions/{execution_id}/cancel\n```\n\n### Retry Execução Falhada\n\n```bash\nPOST /api/executions/{execution_id}/retry\n```\n\n## 💡 Dicas de Uso\n\n### 1. **Estruture bem o input**\n- Use objetos estruturados para inputs complexos\n- Inclua contexto relevante para melhor performance\n- Seja específico sobre o que você quer como output\n\n### 2. **Configure MCPs adequadamente**\n- Firecrawl: Use `onlyMainContent: true` para conteúdo limpo\n- Serper: Especifique localização para resultados relevantes\n- Semrush: Ajuste `display_limit` baseado na necessidade\n\n### 3. **Workflows eficientes**\n- Mapeie inputs/outputs cuidadosamente\n- Use ordem lógica de execução\n- Teste cada agente individualmente primeiro\n\n### 4. **Monitoramento**\n- Sempre monitore execuções longas\n- Use retry para execuções falhadas\n- Analise logs para debugging\n\n### 5. **Performance**\n- Execuções assíncronas para operações longas\n- Use contexto para evitar reprocessamento\n- Configure timeouts adequados",[],[322],{"_102":329,"_95":330,"_151":331,"_153":-5,"_154":136,"_155":332,"_157":158,"_104":333,"_99":156,"_160":298,"_162":334,"_163":22,"_164":-5,"_165":335,"_98":336},"f7312e7a-8dcc-11f0-9899-0224a6c84d84","README.md","f6610948-8dc7-11f0-aaf2-0224a6c84d84",72,["D",1757457190823],"# Anthropic Agent API with MCPs\n\nUma API completa para gerenciar agentes Anthropic com suporte a MCPs (Model Context Protocols) e workflows assíncronos.\n\n## 🚀 Status da API\n\n✅ **API Funcionando**: https://diegoivo--bc8982c68dc811f0833f0224a6c84d84.web.val.run \n✅ **Interface Swagger**: https://diegoivo--bc8982c68dc811f0833f0224a6c84d84.web.val.run/swagger \n✅ **Documentação**: https://diegoivo--bc8982c68dc811f0833f0224a6c84d84.web.val.run/api/docs/swagger.json \n📚 **Guia de Exemplos**: [EXAMPLES.md](./EXAMPLES.md) - Exemplos práticos de como usar os bodies das requests\n\n## Funcionalidades\n\n- 🤖 **Agentes Anthropic**: Integração com Claude Sonnet 4 usando thinking mode\n- 🔗 **MCPs Integrados**: Firecrawl, Serper.dev e Semrush\n- 🔄 **Workflows**: Criação de pipelines onde o output de um agente é input do próximo\n- 📊 **Interface Swagger**: Documentação e teste das APIs\n- 💾 **SQLite**: Persistência de agentes, requisições e workflows\n- ⚡ **Execução Assíncrona**: Suporte para agentes de longa duração\n- 📈 **Monitoramento**: Tracking de status e progresso das execuções\n\n## ⚙️ Configuração das Variáveis de Ambiente\n\n### No Val Town (Recomendado)\n\n1. Acesse seu val no Val Town\n2. Clique no botão **\"Settings\"** ou **\"⚙️\"** \n3. Vá para **\"Environment Variables\"** ou **\"Secrets\"**\n4. Adicione as seguintes variáveis:\n\n```bash\nANTHROPIC_API_KEY=your_anthropic_api_key_here\nFIRECRAWL_API_KEY=your_firecrawl_api_key_here\nSERPER_API_KEY=your_serper_api_key_here\nSEMRUSH_API_KEY=your_semrush_api_key_here\n```\n\n### Onde Conseguir as API Keys\n\n| Serviço | URL | Descrição |\n|---------|-----|-----------|\n| **Anthropic** | https://console.anthropic.com/ | Claude API para agentes inteligentes |\n| **Firecrawl** | https://firecrawl.dev/ | Web scraping e extração de conteúdo |\n| **Serper.dev** | https://serper.dev/ | Google Search API |\n| **Semrush** | https://www.semrush.com/api-documentation/ | SEO e análise de palavras-chave |\n\n## 📚 Estrutura do Projeto\n\n```\n├── backend/\n│ ├── database/\n│ │ ├── migrations.ts # Schema do banco de dados\n│ │ └── queries.ts # Funções de consulta\n│ ├── routes/\n│ │ ├── agents.ts # CRUD de agentes\n│ │ ├── workflows.ts # Gerenciamento de workflows\n│ │ ├── executions.ts # Execuções assíncronas\n│ │ └── swagger.ts # Documentação Swagger\n│ ├── services/\n│ │ ├── anthropic.ts # Integração com Anthropic\n│ │ ├── mcps.ts # Gerenciamento de MCPs\n│ │ └── workflow.ts # Engine de workflows\n│ └── index.ts # Entry point da API\n├── frontend/\n│ └── swagger.html # Interface Swagger\n└── shared/\n └── types.ts # Tipos compartilhados\n```\n\n## 🔧 Uso da API\n\n### 1. Criar um Agente\n\n```bash\ncurl -X POST https://diegoivo--bc8982c68dc811f0833f0224a6c84d84.web.val.run/api/agents \\\n -H \"Content-Type: application/json\" \\\n -d '{\n \"name\": \"Web Research Agent\",\n \"description\": \"Agente especializado em pesquisa web\",\n \"system_prompt\": \"Você é um especialista em pesquisa web. Use as ferramentas disponíveis para encontrar informações relevantes e precisas.\",\n \"mcps\": [\n {\"type\": \"firecrawl\", \"enabled\": true, \"config\": {}},\n {\"type\": \"serper\", \"enabled\": true, \"config\": {}}\n ]\n }'\n```\n\n### 2. Executar um Agente\n\n```bash\ncurl -X POST https://diegoivo--bc8982c68dc811f0833f0224a6c84d84.web.val.run/api/agents/{agent_id}/execute \\\n -H \"Content-Type: application/json\" \\\n -d '{\n \"input\": \"Pesquise sobre as últimas tendências em inteligência artificial para 2024\"\n }'\n```\n\n**Ou com contexto adicional:**\n\n```bash\ncurl -X POST https://diegoivo--bc8982c68dc811f0833f0224a6c84d84.web.val.run/api/agents/{agent_id}/execute \\\n -H \"Content-Type: application/json\" \\\n -d '{\n \"input\": {\n \"content\": \"Pesquise sobre as últimas tendências em inteligência artificial para 2024\",\n \"context\": {\n \"user_id\": \"123\",\n \"session\": \"abc\",\n \"preferences\": [\"tech\", \"AI\"]\n }\n }\n }'\n```\n\n### 3. Criar um Workflow\n\n```bash\ncurl -X POST https://diegoivo--bc8982c68dc811f0833f0224a6c84d84.web.val.run/api/workflows \\\n -H \"Content-Type: application/json\" \\\n -d '{\n \"name\": \"Research and Analysis Workflow\",\n \"description\": \"Pipeline de pesquisa e análise\",\n \"agents\": [\n {\n \"agent_id\": \"agent_1_id\",\n \"order\": 1,\n \"input_mapping\": {},\n \"output_mapping\": {}\n },\n {\n \"agent_id\": \"agent_2_id\", \n \"order\": 2,\n \"input_mapping\": {\"query\": \"results.summary\"},\n \"output_mapping\": {}\n }\n ]\n }'\n```\n\n### 4. Monitorar Execução\n\n```bash\ncurl https://diegoivo--bc8982c68dc811f0833f0224a6c84d84.web.val.run/api/executions/{execution_id}/status\n```\n\n## 🛠️ Endpoints Principais\n\n| Método | Endpoint | Descrição |\n|--------|----------|-----------|\n| `GET` | `/` | Informações da API |\n| `GET` | `/health` | Health check |\n| `GET` | `/swagger` | Interface Swagger |\n| `GET` | `/api/agents` | Listar agentes |\n| `POST` | `/api/agents` | Criar agente |\n| `POST` | `/api/agents/{id}/execute` | Executar agente (assíncrono) |\n| `GET` | `/api/workflows` | Listar workflows |\n| `POST` | `/api/workflows` | Criar workflow |\n| `POST` | `/api/workflows/{id}/execute` | Executar workflow |\n| `GET` | `/api/executions` | Listar execuções |\n| `GET` | `/api/executions/{id}/status` | Status detalhado |\n\n## 🔗 MCPs Suportados\n\n### Firecrawl\n- **Função**: Web scraping e extração de conteúdo\n- **Tool**: `firecrawl_scrape`\n- **Uso**: Extrair conteúdo de páginas web\n\n### Serper.dev \n- **Função**: Pesquisa no Google\n- **Tool**: `serper_search`\n- **Tipos**: search, images, videos, places, maps, news, shopping\n\n### Semrush\n- **Função**: Análise de SEO e palavras-chave\n- **Tool**: `semrush_domain_analysis`\n- **Uso**: Análise de domínios e keywords\n\n## 🧠 Configuração dos Agentes\n\nCada agente é configurado com:\n\n- **Model**: `claude-3-5-sonnet-20241022`\n- **Max Tokens**: 64.000 tokens de output\n- **Context Window**: 1.000.000 tokens de contexto\n- **Thinking Mode**: Ativo por padrão\n- **Budget Tokens**: 32.000 tokens de orçamento\n- **System Prompt**: Personalizável\n- **MCPs**: Configuráveis individualmente\n\n## 📊 Monitoramento\n\nA API oferece monitoramento completo:\n\n- Status em tempo real das execuções\n- Logs detalhados de cada agente\n- Estatísticas de uso e performance\n- Retry automático para execuções falhadas\n- Cancelamento de execuções em andamento\n\n## 🔄 Workflows\n\nOs workflows permitem:\n\n- Execução sequencial de múltiplos agentes\n- Mapeamento de input/output entre agentes\n- Contexto compartilhado entre execuções\n- Monitoramento de progresso\n- Tratamento de erros por etapa\n\n## 🚨 Tratamento de Erros\n\nA API inclui:\n\n- Validação de entrada robusta\n- Tratamento de erros de API externa\n- Logs detalhados para debugging\n- Retry automático para falhas temporárias\n- Status codes HTTP apropriados\n\n## 📈 Performance\n\n- Execução assíncrona para operações longas\n- Otimização de queries SQLite\n- Cache de configurações\n- Monitoramento de tokens utilizados\n- Controle de rate limiting\n\n## 🔐 Segurança\n\n- Variáveis de ambiente para API keys\n- Validação de entrada\n- Sanitização de dados\n- CORS configurado\n- Logs de auditoria",[],[330],"files",[148,167,173,182,193,200,210,220,229,239,245,254,262,272,284,291,303,310,320,328],"counts",{"_341":32,"_342":297,"_343":32,"_344":345},"pulls","branches","forks","envVars",{"_346":278,"_347":32},"count","emptyCount","contents",[350,353,356,359,362],{"_102":149,"_95":150,"_151":152,"_153":-5,"_154":136,"_155":156,"_157":158,"_104":351,"_99":32,"_160":161,"_162":-5,"_163":22,"_164":-5,"_98":352},["D",1757455066717],[150],{"_102":285,"_95":286,"_151":287,"_153":-5,"_154":136,"_155":288,"_157":158,"_104":354,"_99":32,"_160":161,"_162":-5,"_163":22,"_164":-5,"_98":355},["D",1757455417170],[286],{"_102":304,"_95":305,"_151":306,"_153":-5,"_154":136,"_155":307,"_157":158,"_104":357,"_99":32,"_160":161,"_162":-5,"_163":22,"_164":-5,"_98":358},["D",1757455054884],[305],{"_102":321,"_95":322,"_151":323,"_153":-5,"_154":136,"_155":135,"_157":158,"_104":360,"_99":307,"_160":298,"_162":325,"_163":22,"_164":-5,"_98":361},["D",1757457223044],[322],{"_102":329,"_95":330,"_151":331,"_153":-5,"_154":136,"_155":332,"_157":158,"_104":363,"_99":156,"_160":298,"_162":334,"_163":22,"_164":-5,"_98":364},["D",1757457190823],[330],"upstreamProject","upstreamBranch","emptyEnvVars","isNew","flags",[],"routes/_app.x.$handle.$name.environment-variables",{"_93":94,"_95":96,"_100":373,"_344":378},{"_102":103,"_95":96,"_104":374,"_106":-5,"_107":22,"_108":109,"_110":-5,"_111":-5,"_14":112,"_113":-5,"_114":22,"_115":-5,"_116":32,"_16":117,"_118":119,"_120":119,"_121":122,"_123":122,"_124":122,"_125":119,"_126":119,"_23":375,"_130":376,"_133":377,"_145":22},["D",1757454746679],{"_93":94,"_128":129},{"_93":94,"_132":129},{"_99":135,"_95":43,"_102":136,"_137":103,"_138":-5,"_139":-5,"_140":-5,"_141":-5,"_142":-5,"_143":96,"_144":94},[379,385,388,392],{"_380":381,"_104":382,"_383":-5,"_110":72,"_384":22},"key","SEMRUSH_API_KEY",["D",1757456075273],"updatedAt","isEmpty",{"_380":386,"_104":387,"_383":-5,"_110":72,"_384":22},"FIRECRAWL_API_KEY",["D",1757456055855],{"_380":389,"_104":390,"_383":391,"_110":72,"_384":22},"SERPER_API_KEY",["D",1757456042853],["D",1757456079955],{"_380":393,"_104":394,"_383":-5,"_110":72,"_384":22},"ANTHROPIC_API_KEY",["D",1757456032297],"actionData","errors"]

Key	Description
SEMRUSH_API_KEY
FIRECRAWL_API_KEY
SERPER_API_KEY
ANTHROPIC_API_KEY

diegoivo

sdkAnthropic

Environment variables