auspex 0.2.0 → 0.3.0

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "auspex",
-  "version": "0.2.0",
+  "version": "0.3.0",
   "description": "AI-powered browser agent with tiered scraping — HTTP/Cheerio first, Playwright as fallback",
   "type": "module",
   "main": "./dist/index.js",
@@ -20,7 +20,9 @@
     "dev": "tsx src/index.ts",
     "example": "tsx examples/basic.ts",
     "example:scraper": "tsx examples/scraper.ts",
-    "install:browser": "playwright install chromium"
+    "install:browser": "playwright install chromium",
+    "test": "vitest run",
+    "test:watch": "vitest"
   },
   "keywords": [
     "auspex",
@@ -42,9 +44,12 @@
     "openai": "^4.77.0",
     "playwright": "^1.49.0",
     "playwright-core": "^1.49.0",
+    "playwright-extra": "^4.3.6",
+    "puppeteer-extra-plugin-stealth": "^2.11.2",
     "turndown": "^7.2.2",
     "turndown-plugin-gfm": "^1.0.2",
-    "zod": "^3.24.0"
+    "zod": "^3.24.0",
+    "zod-to-json-schema": "^3.25.1"
   },
   "devDependencies": {
     "@types/jsdom": "^21.1.7",
@@ -52,6 +57,7 @@
     "@types/turndown": "^5.0.6",
     "dotenv": "^16.4.0",
     "tsx": "^4.19.0",
-    "typescript": "^5.7.0"
+    "typescript": "^5.7.0",
+    "vitest": "^4.0.18"
   }
 }

package/readme.md

@@ -18,6 +18,8 @@ Framework de browser automation alimentado por LLM. Voce fornece uma **URL** e u
 - [Relatorio de Execucao](#relatorio-de-execucao)
 - [Parametros da LLM](#parametros-da-llm)
 - [Providers de LLM compativeis](#providers-de-llm-compativeis)
+- [Eventos](#eventos)
+- [Browser Pool](#browser-pool)
 - [Acoes do Agent](#acoes-do-agent)
 - [Seguranca](#seguranca)
 - [Monitoramento — Tokens e Memoria](#monitoramento--tokens-e-memoria)
@@ -126,6 +128,20 @@ const r2 = await agent.run({
 await agent.close(); // limpa tudo no final
 ```
 
+### Construtor com pool (opcional)
+
+Voce pode passar um `BrowserPool` como segundo argumento para compartilhar instancias de browser entre multiplos agents (ex.: workers concorrentes). Sem pool, o agent usa um unico browser por instancia.
+
+```typescript
+import { Auspex, BrowserPool } from "auspex";
+
+const pool = new BrowserPool({ maxSize: 3 });
+const agent = new Auspex({ llmApiKey: "sk-..." }, pool);
+const result = await agent.run({ url, prompt });
+await agent.close(); // apenas desvincula do pool; o pool continua ativo
+// pool.close() quando nao for mais usar
+```
+
 ---
 
 ## AgentConfig — Configuracao completa
@@ -154,6 +170,22 @@ new Auspex({
   // ──── Seguranca ──────────────────────────────────
   allowedDomains: ["example.com"], // se definido, SO permite esses dominios
   blockedDomains: ["evil.com"],    // dominios bloqueados explicitamente
+
+  // ──── Browser e rede ────────────────────────────
+  gotoTimeoutMs: 15000,            // timeout do page.goto (default: 15s)
+  proxy: { server: "http://...", username?: "...", password?: "..." },
+  cookies: [{ name, value, domain?, path?, ... }],  // cookies injetados no context
+  extraHeaders: { "Accept-Language": "pt-BR" },     // headers HTTP do context
+
+  // ──── Loop e orcamento ──────────────────────────
+  actionDelayMs: 500,              // delay entre iteracoes (ms, default: 500)
+  maxTotalTokens: 0,             // orcamento total de tokens (0 = ilimitado)
+
+  // ──── Log e vision ───────────────────────────────
+  log: false,                     // gravar log em arquivo por execucao (./logs/)
+  logDir: "logs",                 // diretorio dos logs
+  vision: false,                  // fallback com screenshot apos falhas (modelo com vision)
+  screenshotQuality: 75,           // qualidade JPEG 1-100 (vision, default: 75)
 });
 ```
 
@@ -172,8 +204,18 @@ new Auspex({
 | `maxIterations` | `number` | Nao | `30` | Max iteracoes do agent loop |
 | `timeoutMs` | `number` | Nao | `120000` | Timeout total da execucao (ms) |
 | `maxWaitMs` | `number` | Nao | `5000` | Max ms para acao `wait` |
+| `gotoTimeoutMs` | `number` | Nao | `15000` | Timeout do page.goto (ms) |
 | `allowedDomains` | `string[]` | Nao | — | Whitelist de dominios permitidos |
 | `blockedDomains` | `string[]` | Nao | — | Blacklist de dominios bloqueados |
+| `actionDelayMs` | `number` | Nao | `500` | Delay entre iteracoes (ms) |
+| `maxTotalTokens` | `number` | Nao | `0` | Orcamento total de tokens (0 = ilimitado) |
+| `proxy` | `ProxyConfig` | Nao | — | Proxy para browser e requests |
+| `cookies` | `CookieParam[]` | Nao | — | Cookies injetados no context |
+| `extraHeaders` | `Record<string, string>` | Nao | — | Headers HTTP do context |
+| `log` | `boolean` | Nao | `false` | Gravar log em arquivo por run |
+| `logDir` | `string` | Nao | `"logs"` | Diretorio dos arquivos de log |
+| `vision` | `boolean` | Nao | `false` | Fallback com screenshot apos falhas (modelo vision) |
+| `screenshotQuality` | `number` | Nao | `75` | Qualidade JPEG 1-100 para screenshots |
 
 ---
 
@@ -185,12 +227,27 @@ Opcoes passadas para `agent.run(options)`:
 |-----------|------|:-----------:|-----------|
 | `url` | `string` | Sim | URL inicial para o agent navegar |
 | `prompt` | `string` | Sim | Instrucao em linguagem natural |
+| `maxIterations` | `number` | Nao | Override de maxIterations para este run |
+| `timeoutMs` | `number` | Nao | Override de timeoutMs para este run |
+| `actionDelayMs` | `number` | Nao | Override de actionDelayMs para este run |
+| `signal` | `AbortSignal` | Nao | AbortSignal para cancelar o run |
+| `schema` | `ZodType<T>` | Nao | Schema Zod: retorno tipado em `data` (T \| null) |
+| `vision` | `boolean` | Nao | Override do fallback com screenshot |
 
 ```typescript
 const result = await agent.run({
   url: "https://example.com",
   prompt: "Qual o titulo desta pagina?",
 });
+
+// Com schema Zod — data fica tipado
+const schema = z.object({ title: z.string(), price: z.number() });
+const result = await agent.run({
+  url: "https://shop.example.com",
+  prompt: "Extraia titulo e preco do produto.",
+  schema,
+});
+// result.data: { title: string; price: number } | null
 ```
 
 ---
@@ -201,9 +258,9 @@ O `agent.run()` retorna um objeto `AgentResult` com tudo que aconteceu:
 
 ```typescript
 interface AgentResult {
-  status: "done" | "max_iterations" | "error" | "timeout";
+  status: "done" | "max_iterations" | "error" | "timeout" | "aborted";
   tier: "http" | "playwright";    // metodo de scraping utilizado
-  data: string | null;            // resultado retornado pelo agent (texto)
+  data: string | null;            // resultado (texto). Com run({ schema }), pode ser objeto tipado
   report: string;                 // relatorio formatado legivel
   durationMs: number;             // duracao total da execucao em ms
   actions: ActionRecord[];        // historico de todas as acoes executadas
@@ -220,6 +277,7 @@ interface AgentResult {
 | `"done"` | Tarefa concluida com sucesso. `data` contem o resultado. |
 | `"max_iterations"` | Atingiu o limite de iteracoes sem concluir. |
 | `"timeout"` | Tempo limite excedido (`timeoutMs`). |
+| `"aborted"` | Cancelado pelo chamador (AbortSignal). |
 | `"error"` | Erro durante execucao. Ver `error` para detalhes. |
 
 ### Tier
@@ -272,31 +330,32 @@ const result = await agent.run({ url, prompt });
 console.log(result.report);
 ```
 
-Exemplo de saida (tier HTTP):
+Exemplo de saida (tier HTTP). O relatorio eh gerado em ingles:
 
 ```
 ═══════════════════════════════════════════
-  RELATORIO DE EXECUCAO — auspex
+  EXECUTION REPORT — auspex
 ═══════════════════════════════════════════
 
-URL    : https://news.ycombinator.com
-Prompt : Retorne o titulo do primeiro artigo.
-Status : Tarefa concluida com sucesso.
-Metodo : 🟢 HTTP/Cheerio (sem browser)
-Duracao: 1.2s
+  URL     : https://news.ycombinator.com
+  Prompt  : Retorne o titulo do primeiro artigo.
+  Status  : Task completed successfully.
+  Method  : HTTP/Cheerio (no browser — static page)
+  Duration: 1.2s
 
 ───────────────────────────────────────────
-  RESULTADO
+  RESULT
 ───────────────────────────────────────────
 
   Show HN: My weekend project
 
 ───────────────────────────────────────────
-  CONSUMO
+  RESOURCE USAGE
 ───────────────────────────────────────────
 
-  LLM   : 1 chamada | 1820 tokens (1650 prompt + 170 completion)
-  Memoria: Node.js 45.2 MB | Chromium: nao utilizado
+  LLM    : 1 call(s) | 1820 tokens
+           > 1650 prompt + 170 completion
+  RAM    : Node.js 45.2 MB  |  Browser: not used
 
 ═══════════════════════════════════════════
 ```
@@ -305,37 +364,38 @@ Exemplo de saida (tier Playwright):
 
 ```
 ═══════════════════════════════════════════
-  RELATORIO DE EXECUCAO — auspex
+  EXECUTION REPORT — auspex
 ═══════════════════════════════════════════
 
-URL    : https://app.exemplo.com
-Prompt : Faca login e retorne o saldo da conta.
-Status : Tarefa concluida com sucesso.
-Metodo : 🟡 Playwright Chromium
-Duracao: 12.4s
+  URL     : https://app.exemplo.com
+  Prompt  : Faca login e retorne o saldo da conta.
+  Status  : Task completed successfully.
+  Method  : Playwright Chromium (full browser — JS required)
+  Duration: 12.4s
 
 ───────────────────────────────────────────
-  PASSO A PASSO
+  STEP BY STEP
 ───────────────────────────────────────────
 
-  1. Clicou no elemento "input[name='email']"
-  2. Digitou "user@email.com" em "input[name='email']"
-  3. Digitou "••••••••" em "input[name='password']"
-  4. Clicou no elemento "button[type='submit']"
-  5. Finalizou com resultado
+  1. Clicked element "input[name='email']"
+  2. Typed "user@email.com" into "input[name='email']"
+  3. Typed "••••••••" into "input[name='password']"
+  4. Clicked element "button[type='submit']"
+  5. Finished with result
 
 ───────────────────────────────────────────
-  RESULTADO
+  RESULT
 ───────────────────────────────────────────
 
   Saldo: R$ 1.234,56
 
 ───────────────────────────────────────────
-  CONSUMO
+  RESOURCE USAGE
 ───────────────────────────────────────────
 
-  LLM   : 5 chamadas | 9430 tokens (8100 prompt + 1330 completion)
-  Memoria: Node.js 67.0 MB | Chromium pico: 412.3 MB
+  LLM    : 5 call(s) | 9430 tokens
+           > 8100 prompt + 1330 completion
+  RAM    : Node.js 67.0 MB  |  Chromium peak 412.3 MB
 
 ═══════════════════════════════════════════
 ```
@@ -443,14 +503,72 @@ O LLM so pode executar acoes de uma **whitelist rigorosa**. Qualquer coisa fora
 
 | Acao | Formato JSON | Descricao |
 |------|-------------|-----------|
-| **click** | `{"type":"click","selector":"#btn"}` | Clica em um elemento via CSS selector |
+| **click** | `{"type":"click","selector":"#btn"}` | Clica em um elemento (CSS ou role=button[name="..."] ) |
 | **type** | `{"type":"type","selector":"input[name='q']","text":"busca"}` | Digita texto em um campo (max 1000 chars) |
+| **select** | `{"type":"select","selector":"select#country","value":"br"}` | Seleciona opcao em `<select>` (value = option value) |
+| **pressKey** | `{"type":"pressKey","key":"Enter"}` | Tecla: Enter, Tab, Escape, Backspace, ArrowUp/Down, etc. |
+| **hover** | `{"type":"hover","selector":"#menu"}` | Passa o mouse sobre o elemento (menus, tooltips) |
 | **goto** | `{"type":"goto","url":"https://..."}` | Navega para uma URL (passa por validacao anti-SSRF) |
 | **wait** | `{"type":"wait","ms":2000}` | Espera N milissegundos (max 5000ms) |
-| **scroll** | `{"type":"scroll","direction":"down"}` | Scroll para cima ou para baixo |
-| **done** | `{"type":"done","result":"titulo da pagina"}` | Finaliza e retorna o resultado |
+| **scroll** | `{"type":"scroll","direction":"down","amount":500}` | Scroll (amount opcional, default 500px) |
+| **done** | `{"type":"done","result":"..."}` | Finaliza e retorna o resultado (max 50k chars) |
+
+Selectors podem ser **CSS** ou **role-based** (ex.: `role=button[name="Submit"]`) quando a Accessibility Tree esta no snapshot.
+
+---
+
+## Eventos
+
+O `Auspex` estende `EventEmitter`. Voce pode escutar eventos por run:
+
+| Evento | Argumentos | Descricao |
+|--------|------------|-----------|
+| `tier` | `(tier: AgentTier)` | Indica se o run usou HTTP ou Playwright |
+| `iteration` | `(iteration: number, snapshot: PageSnapshot)` | Apos cada snapshot no loop |
+| `action` | `(action: AgentAction, iteration: number)` | Antes de executar cada acao |
+| `error` | `(error: Error)` | Em caso de erro na execucao (se houver listener) |
+| `done` | `(result: AgentResult)` | Ao finalizar o run (sucesso ou nao) |
+
+```typescript
+agent.on("tier", (tier) => console.log("[tier]", tier));
+agent.on("iteration", (i, snapshot) => console.log(`[iter ${i}]`, snapshot.url));
+agent.on("action", (action, i) => console.log(`[action ${i}]`, action.type));
+agent.on("done", (result) => console.log("Done:", result.status));
+
+const result = await agent.run({ url, prompt });
+```
+
+---
+
+## Browser Pool
+
+A classe `BrowserPool` gerencia um conjunto de browsers Playwright reutilizaveis. Util para limitar recursos quando varios agents rodam em paralelo.
+
+```typescript
+import { BrowserPool } from "auspex";
+
+const pool = new BrowserPool({
+  maxSize: 3,
+  acquireTimeoutMs: 30_000,
+  launchOptions: { headless: true },
+});
 
-### Acoes BLOQUEADAS
+const browser = await pool.acquire();
+// ... usar browser (newContext, newPage, etc.)
+pool.release(browser);
+
+await pool.close(); // fecha todos os browsers
+```
+
+| Opcao | Tipo | Default | Descricao |
+|-------|------|---------|-----------|
+| `maxSize` | `number` | `3` | Maximo de instancias de browser |
+| `acquireTimeoutMs` | `number` | `30000` | Timeout ao esperar um browser livre |
+| `launchOptions` | `LaunchOptions` | headless + stealth args | Opcoes do Playwright |
+
+---
+
+## Acoes BLOQUEADAS
 
 O framework **nao permite** nenhuma forma de:
 
@@ -473,7 +591,7 @@ Toda URL (inicial e durante navegacao) passa por validacao rigorosa:
 - **IPs privados**: `127.0.0.0/8`, `10.0.0.0/8`, `192.168.0.0/16`, `172.16.0.0/12`
 - **Cloud metadata**: `169.254.169.254` (AWS/GCP metadata endpoint)
 - **Localhost**: `localhost`, `[::1]`
-- **DNS rebinding**: resolve o hostname antes de navegar — detecta dominios publicos apontando para IPs privados
+- **DNS rebinding**: resolve o hostname antes de navegar — detecta dominios publicos apontando para IPs privados. Falha de DNS rejeita a URL (fail closed).
 
 ### Whitelist de acoes
 
@@ -486,6 +604,8 @@ Selectors CSS sao verificados contra padroes maliciosos antes de qualquer intera
 - `javascript:` — bloqueado
 - `on*=` (event handlers como `onclick=`) — bloqueado
 - `<script>` — bloqueado
+- `data:` — bloqueado
+- Tamanho maximo de 500 caracteres (protecao DoS)
 - Strings vazias ou so espacos — bloqueadas
 
 ### Anti-prompt injection
@@ -613,6 +733,9 @@ switch (result.status) {
   case "timeout":
     console.warn("Timeout — aumente timeoutMs ou simplifique a tarefa");
     break;
+  case "aborted":
+    console.warn("Cancelado (AbortSignal)");
+    break;
   case "error":
     console.error("Erro:", result.error);
     break;
@@ -632,7 +755,8 @@ Criar um `Auspex` uma vez e chamar `run()` multiplas vezes eh mais eficiente do
 - **Sem file upload**: a acao `type` preenche campos de texto, mas nao faz upload de arquivos.
 - **Dependencia de selectors CSS**: a qualidade da automacao depende da capacidade do LLM de identificar selectors corretos a partir do snapshot textual.
 - **Snapshot limitado**: captura ate 25 links, 5 formularios e 3500 chars de texto por pagina. Paginas muito grandes podem ter elementos nao capturados.
-- **JSON mode obrigatorio**: o provider LLM deve suportar `response_format: { type: "json_object" }`.
+- **Vision**: ao usar `vision: true`, o modelo deve suportar entrada de imagem (ex.: gpt-4o, gpt-4o-mini). Screenshot so eh enviado apos falhas consecutivas (fallback).
+- **JSON mode obrigatorio**: o provider LLM deve suportar `response_format: { type: "json_object" }` (exceto em chamadas com screenshot, quando pode ser omitido).
 
 ---
 
@@ -694,6 +818,11 @@ src/
     loop.ts                 # Core loop: snapshot -> LLM -> validar -> executar
     actions.ts              # Parser e validador de acoes do LLM
     report.ts               # Gerador de relatorio de execucao
+    logger.ts               # Log por run (arquivo em logDir)
+  browser/
+    pool.ts                 # BrowserPool — pool de browsers reutilizaveis
+  llm/
+    vision-models.ts        # Whitelist de modelos com suporte a vision
   scraper/
     index.ts                # Scraper — fallback HTTP -> Stealth -> Browser
     tiers/
@@ -777,6 +906,7 @@ LLM_API_KEY=sk-your-key-here
 ```typescript
 import {
   Auspex,
+  BrowserPool,
   type AgentConfig,
   type AgentResult,
   type AgentAction,
@@ -790,6 +920,10 @@ import {
   type SnapshotLink,
   type SnapshotForm,
   type SnapshotInput,
+  type ProxyConfig,
+  type CookieParam,
+  type AuspexEvents,
+  type BrowserPoolOptions,
   UrlValidationError,
   ActionValidationError,
 } from "auspex";

package/dist/agent/actions.d.ts

@@ -1,5 +0,0 @@
-import type { AgentAction } from "../types.js";
-import { type UrlValidationOptions } from "../security/url-validator.js";
-export declare function parseAndValidateAction(raw: unknown, urlOptions: UrlValidationOptions): Promise<AgentAction>;
-export declare function formatActionForHistory(action: AgentAction, iteration: number): string;
-//# sourceMappingURL=actions.d.ts.map

package/dist/agent/actions.d.ts.map

@@ -1 +0,0 @@
-{"version":3,"file":"actions.d.ts","sourceRoot":"","sources":["../../src/agent/actions.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,WAAW,EAAE,MAAM,aAAa,CAAC;AAE/C,OAAO,EAAe,KAAK,oBAAoB,EAAE,MAAM,8BAA8B,CAAC;AAEtF,wBAAsB,sBAAsB,CAC1C,GAAG,EAAE,OAAO,EACZ,UAAU,EAAE,oBAAoB,GAC/B,OAAO,CAAC,WAAW,CAAC,CAQtB;AAED,wBAAgB,sBAAsB,CAAC,MAAM,EAAE,WAAW,EAAE,SAAS,EAAE,MAAM,GAAG,MAAM,CAerF"}

package/dist/agent/actions.js

@@ -1,26 +0,0 @@
-import { validateAction } from "../security/action-validator.js";
-import { validateUrl } from "../security/url-validator.js";
-export async function parseAndValidateAction(raw, urlOptions) {
-    const action = validateAction(raw);
-    if (action.type === "goto") {
-        await validateUrl(action.url, urlOptions);
-    }
-    return action;
-}
-export function formatActionForHistory(action, iteration) {
-    switch (action.type) {
-        case "click":
-            return `[${iteration}] click "${action.selector}"`;
-        case "type":
-            return `[${iteration}] type "${action.text}" into "${action.selector}"`;
-        case "goto":
-            return `[${iteration}] navigate to ${action.url}`;
-        case "wait":
-            return `[${iteration}] wait ${action.ms}ms`;
-        case "scroll":
-            return `[${iteration}] scroll ${action.direction}`;
-        case "done":
-            return `[${iteration}] done: ${action.result}`;
-    }
-}
-//# sourceMappingURL=actions.js.map

package/dist/agent/actions.js.map

@@ -1 +0,0 @@
-{"version":3,"file":"actions.js","sourceRoot":"","sources":["../../src/agent/actions.ts"],"names":[],"mappings":"AACA,OAAO,EAAE,cAAc,EAAyB,MAAM,iCAAiC,CAAC;AACxF,OAAO,EAAE,WAAW,EAA6B,MAAM,8BAA8B,CAAC;AAEtF,MAAM,CAAC,KAAK,UAAU,sBAAsB,CAC1C,GAAY,EACZ,UAAgC;IAEhC,MAAM,MAAM,GAAG,cAAc,CAAC,GAAG,CAAC,CAAC;IAEnC,IAAI,MAAM,CAAC,IAAI,KAAK,MAAM,EAAE,CAAC;QAC3B,MAAM,WAAW,CAAC,MAAM,CAAC,GAAG,EAAE,UAAU,CAAC,CAAC;IAC5C,CAAC;IAED,OAAO,MAAqB,CAAC;AAC/B,CAAC;AAED,MAAM,UAAU,sBAAsB,CAAC,MAAmB,EAAE,SAAiB;IAC3E,QAAQ,MAAM,CAAC,IAAI,EAAE,CAAC;QACpB,KAAK,OAAO;YACV,OAAO,IAAI,SAAS,YAAY,MAAM,CAAC,QAAQ,GAAG,CAAC;QACrD,KAAK,MAAM;YACT,OAAO,IAAI,SAAS,WAAW,MAAM,CAAC,IAAI,WAAW,MAAM,CAAC,QAAQ,GAAG,CAAC;QAC1E,KAAK,MAAM;YACT,OAAO,IAAI,SAAS,iBAAiB,MAAM,CAAC,GAAG,EAAE,CAAC;QACpD,KAAK,MAAM;YACT,OAAO,IAAI,SAAS,UAAU,MAAM,CAAC,EAAE,IAAI,CAAC;QAC9C,KAAK,QAAQ;YACX,OAAO,IAAI,SAAS,YAAY,MAAM,CAAC,SAAS,EAAE,CAAC;QACrD,KAAK,MAAM;YACT,OAAO,IAAI,SAAS,WAAW,MAAM,CAAC,MAAM,EAAE,CAAC;IACnD,CAAC;AACH,CAAC"}

package/dist/agent/agent.d.ts

@@ -1,12 +0,0 @@
-import type { AgentConfig, AgentResult, RunOptions } from "../types.js";
-export declare class Auspex {
-    private config;
-    private browser;
-    constructor(config: AgentConfig);
-    private getChromiumPids;
-    private makeChromiumMemoryTracker;
-    private ensureBrowser;
-    run(options: RunOptions): Promise<AgentResult>;
-    close(): Promise<void>;
-}
-//# sourceMappingURL=agent.d.ts.map

package/dist/agent/agent.d.ts.map

@@ -1 +0,0 @@
-{"version":3,"file":"agent.d.ts","sourceRoot":"","sources":["../../src/agent/agent.ts"],"names":[],"mappings":"AAGA,OAAO,KAAK,EAAE,WAAW,EAAE,WAAW,EAAE,UAAU,EAAE,MAAM,aAAa,CAAC;AAOxE,qBAAa,MAAM;IACjB,OAAO,CAAC,MAAM,CAAkB;IAChC,OAAO,CAAC,OAAO,CAAwB;gBAE3B,MAAM,EAAE,WAAW;IAM/B,OAAO,CAAC,eAAe;IAcvB,OAAO,CAAC,yBAAyB;YA4BnB,aAAa;IAYrB,GAAG,CAAC,OAAO,EAAE,UAAU,GAAG,OAAO,CAAC,WAAW,CAAC;IA4F9C,KAAK,IAAI,OAAO,CAAC,IAAI,CAAC;CAM7B"}

package/dist/agent/agent.js

@@ -1,159 +0,0 @@
-import { execSync } from "node:child_process";
-import { chromium } from "playwright";
-import { gotScraping } from "got-scraping";
-import { agentConfigSchema, runOptionsSchema } from "../config/schema.js";
-import { validateUrl } from "../security/url-validator.js";
-import { snapshotFromHtml } from "../browser/snapshot.js";
-import { runStaticLoop, runAgentLoop } from "./loop.js";
-import { generateReport } from "./report.js";
-export class Auspex {
-    config;
-    browser = null;
-    constructor(config) {
-        this.config = agentConfigSchema.parse(config);
-    }
-    // ── Snapshot de PIDs do Chromium antes do launch ─────────────────────────
-    getChromiumPids() {
-        try {
-            const out = execSync("ps aux | grep -i chromium | grep -v grep | awk '{print $2}'", {
-                encoding: "utf-8",
-                timeout: 2_000,
-            });
-            return new Set(out.trim().split("\n").filter(Boolean));
-        }
-        catch {
-            return new Set();
-        }
-    }
-    // ── Soma RSS de todos os processos Chromium novos (KB) ────────────────────
-    makeChromiumMemoryTracker(pidsBeforeLaunch) {
-        return () => {
-            try {
-                const out = execSync("ps aux | grep -i chromium | grep -v grep | awk '{print $2}'", {
-                    encoding: "utf-8",
-                    timeout: 2_000,
-                });
-                const newPids = out.trim().split("\n").filter(p => p && !pidsBeforeLaunch.has(p));
-                if (newPids.length === 0)
-                    return 0;
-                // Soma o RSS de todos os processos Chromium lançados por este agente
-                const pidsArg = newPids.join(",");
-                const rssOut = execSync(`ps -o rss= -p ${pidsArg}`, {
-                    encoding: "utf-8",
-                    timeout: 2_000,
-                });
-                return rssOut
-                    .trim()
-                    .split("\n")
-                    .reduce((sum, line) => sum + (parseInt(line.trim(), 10) || 0), 0);
-            }
-            catch {
-                return 0;
-            }
-        };
-    }
-    // ── Garante que o browser Playwright está rodando ─────────────────────────
-    async ensureBrowser() {
-        if (!this.browser || !this.browser.isConnected()) {
-            this.browser = await chromium.launch({
-                headless: true,
-                args: ["--disable-blink-features=AutomationControlled"],
-            });
-        }
-        return this.browser;
-    }
-    // ── Execução principal ────────────────────────────────────────────────────
-    async run(options) {
-        const { url, prompt } = runOptionsSchema.parse(options);
-        const validUrl = await validateUrl(url, {
-            allowedDomains: this.config.allowedDomains,
-            blockedDomains: this.config.blockedDomains,
-        });
-        // ── 1. Tenta HTTP/Cheerio primeiro (sem browser) ──────────────────────
-        //
-        // Se o HTML da página já tiver conteúdo suficiente E o LLM conseguir
-        // responder com "done" na primeira tentativa → economizamos o Playwright
-        // (zero processos externos, ~100-500ms vs ~5-30s).
-        //
-        try {
-            const response = await gotScraping({
-                url: validUrl,
-                method: "GET",
-                headerGeneratorOptions: {
-                    browsers: [{ name: "chrome", minVersion: 120 }],
-                    operatingSystems: ["macos", "windows"],
-                    devices: ["desktop"],
-                    locales: ["pt-BR", "pt", "en-US"],
-                },
-                timeout: { request: Math.min(this.config.timeoutMs, 10_000) },
-                followRedirect: true,
-                maxRedirects: 10,
-                throwHttpErrors: false,
-                decompress: true,
-            });
-            const { body: html, statusCode: status } = response;
-            if (html && status < 400) {
-                const snapshot = snapshotFromHtml(html, validUrl);
-                // Só tenta o loop estático se houver conteúdo mínimo (evita SPA vazia)
-                if (snapshot.text.length > 200) {
-                    const staticResult = await runStaticLoop(snapshot, validUrl, prompt, this.config);
-                    if (staticResult)
-                        return staticResult; // ✅ resolvido sem browser
-                }
-            }
-        }
-        catch {
-            // HTTP falhou (timeout, SSL, rede) → vai direto para o Playwright
-        }
-        // ── 2. Playwright (fallback — página precisa de JS ou de interação) ───
-        const pidsBefore = this.getChromiumPids();
-        const browser = await this.ensureBrowser();
-        const getMemoryKb = this.makeChromiumMemoryTracker(pidsBefore);
-        let page = null;
-        try {
-            const context = await browser.newContext({
-                viewport: { width: 1280, height: 720 },
-                locale: "pt-BR",
-                timezoneId: "America/Sao_Paulo",
-            });
-            context.addInitScript(() => {
-                Object.defineProperty(navigator, "webdriver", { get: () => false, configurable: true });
-                window.chrome = { runtime: {} };
-            });
-            page = await context.newPage();
-            const gotoTimeout = this.config.gotoTimeoutMs ?? 15_000;
-            await page.goto(validUrl, { waitUntil: "domcontentloaded", timeout: gotoTimeout });
-            return await runAgentLoop(page, validUrl, prompt, this.config, getMemoryKb);
-        }
-        catch (err) {
-            const errorMsg = err instanceof Error ? err.message : String(err);
-            const result = {
-                status: "error",
-                tier: "playwright",
-                data: null,
-                report: "",
-                durationMs: 0,
-                actions: [],
-                usage: { promptTokens: 0, completionTokens: 0, totalTokens: 0, calls: 0 },
-                memory: { browserPeakRssKb: 0, nodeHeapUsedMb: 0 },
-                error: errorMsg,
-            };
-            result.report = generateReport(result, validUrl, prompt);
-            return result;
-        }
-        finally {
-            if (page) {
-                const context = page.context();
-                await page.close().catch(() => { });
-                await context.close().catch(() => { });
-            }
-        }
-    }
-    async close() {
-        if (this.browser) {
-            await this.browser.close().catch(() => { });
-            this.browser = null;
-        }
-    }
-}
-//# sourceMappingURL=agent.js.map

package/dist/agent/agent.js.map

@@ -1 +0,0 @@
-{"version":3,"file":"agent.js","sourceRoot":"","sources":["../../src/agent/agent.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,QAAQ,EAAE,MAAM,oBAAoB,CAAC;AAC9C,OAAO,EAAE,QAAQ,EAA2B,MAAM,YAAY,CAAC;AAC/D,OAAO,EAAE,WAAW,EAAE,MAAM,cAAc,CAAC;AAE3C,OAAO,EAAE,iBAAiB,EAAE,gBAAgB,EAAwB,MAAM,qBAAqB,CAAC;AAChG,OAAO,EAAE,WAAW,EAAE,MAAM,8BAA8B,CAAC;AAC3D,OAAO,EAAE,gBAAgB,EAAE,MAAM,wBAAwB,CAAC;AAC1D,OAAO,EAAE,aAAa,EAAE,YAAY,EAAE,MAAM,WAAW,CAAC;AACxD,OAAO,EAAE,cAAc,EAAE,MAAM,aAAa,CAAC;AAE7C,MAAM,OAAO,MAAM;IACT,MAAM,CAAkB;IACxB,OAAO,GAAmB,IAAI,CAAC;IAEvC,YAAY,MAAmB;QAC7B,IAAI,CAAC,MAAM,GAAG,iBAAiB,CAAC,KAAK,CAAC,MAAM,CAAC,CAAC;IAChD,CAAC;IAED,4EAA4E;IAEpE,eAAe;QACrB,IAAI,CAAC;YACH,MAAM,GAAG,GAAG,QAAQ,CAAC,6DAA6D,EAAE;gBAClF,QAAQ,EAAE,OAAO;gBACjB,OAAO,EAAE,KAAK;aACf,CAAC,CAAC;YACH,OAAO,IAAI,GAAG,CAAC,GAAG,CAAC,IAAI,EAAE,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC,MAAM,CAAC,OAAO,CAAC,CAAC,CAAC;QACzD,CAAC;QAAC,MAAM,CAAC;YACP,OAAO,IAAI,GAAG,EAAE,CAAC;QACnB,CAAC;IACH,CAAC;IAED,6EAA6E;IAErE,yBAAyB,CAAC,gBAA6B;QAC7D,OAAO,GAAG,EAAE;YACV,IAAI,CAAC;gBACH,MAAM,GAAG,GAAG,QAAQ,CAAC,6DAA6D,EAAE;oBAClF,QAAQ,EAAE,OAAO;oBACjB,OAAO,EAAE,KAAK;iBACf,CAAC,CAAC;gBACH,MAAM,OAAO,GAAG,GAAG,CAAC,IAAI,EAAE,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,IAAI,CAAC,gBAAgB,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC;gBAClF,IAAI,OAAO,CAAC,MAAM,KAAK,CAAC;oBAAE,OAAO,CAAC,CAAC;gBAEnC,qEAAqE;gBACrE,MAAM,OAAO,GAAG,OAAO,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC;gBAClC,MAAM,MAAM,GAAG,QAAQ,CAAC,iBAAiB,OAAO,EAAE,EAAE;oBAClD,QAAQ,EAAE,OAAO;oBACjB,OAAO,EAAE,KAAK;iBACf,CAAC,CAAC;gBACH,OAAO,MAAM;qBACV,IAAI,EAAE;qBACN,KAAK,CAAC,IAAI,CAAC;qBACX,MAAM,CAAC,CAAC,GAAG,EAAE,IAAI,EAAE,EAAE,CAAC,GAAG,GAAG,CAAC,QAAQ,CAAC,IAAI,CAAC,IAAI,EAAE,EAAE,EAAE,CAAC,IAAI,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC;YACtE,CAAC;YAAC,MAAM,CAAC;gBACP,OAAO,CAAC,CAAC;YACX,CAAC;QACH,CAAC,CAAC;IACJ,CAAC;IAED,6EAA6E;IAErE,KAAK,CAAC,aAAa;QACzB,IAAI,CAAC,IAAI,CAAC,OAAO,IAAI,CAAC,IAAI,CAAC,OAAO,CAAC,WAAW,EAAE,EAAE,CAAC;YACjD,IAAI,CAAC,OAAO,GAAG,MAAM,QAAQ,CAAC,MAAM,CAAC;gBACnC,QAAQ,EAAE,IAAI;gBACd,IAAI,EAAE,CAAC,+CAA+C,CAAC;aACxD,CAAC,CAAC;QACL,CAAC;QACD,OAAO,IAAI,CAAC,OAAO,CAAC;IACtB,CAAC;IAED,6EAA6E;IAE7E,KAAK,CAAC,GAAG,CAAC,OAAmB;QAC3B,MAAM,EAAE,GAAG,EAAE,MAAM,EAAE,GAAG,gBAAgB,CAAC,KAAK,CAAC,OAAO,CAAC,CAAC;QAExD,MAAM,QAAQ,GAAG,MAAM,WAAW,CAAC,GAAG,EAAE;YACtC,cAAc,EAAE,IAAI,CAAC,MAAM,CAAC,cAAc;YAC1C,cAAc,EAAE,IAAI,CAAC,MAAM,CAAC,cAAc;SAC3C,CAAC,CAAC;QAEH,yEAAyE;QACzE,EAAE;QACF,qEAAqE;QACrE,yEAAyE;QACzE,mDAAmD;QACnD,EAAE;QACF,IAAI,CAAC;YACH,MAAM,QAAQ,GAAG,MAAM,WAAW,CAAC;gBACjC,GAAG,EAAE,QAAQ;gBACb,MAAM,EAAE,KAAK;gBACb,sBAAsB,EAAE;oBACtB,QAAQ,EAAE,CAAC,EAAE,IAAI,EAAE,QAAQ,EAAE,UAAU,EAAE,GAAG,EAAE,CAAC;oBAC/C,gBAAgB,EAAE,CAAC,OAAO,EAAE,SAAS,CAAC;oBACtC,OAAO,EAAE,CAAC,SAAS,CAAC;oBACpB,OAAO,EAAE,CAAC,OAAO,EAAE,IAAI,EAAE,OAAO,CAAC;iBAClC;gBACD,OAAO,EAAE,EAAE,OAAO,EAAE,IAAI,CAAC,GAAG,CAAC,IAAI,CAAC,MAAM,CAAC,SAAS,EAAE,MAAM,CAAC,EAAE;gBAC7D,cAAc,EAAE,IAAI;gBACpB,YAAY,EAAE,EAAE;gBAChB,eAAe,EAAE,KAAK;gBACtB,UAAU,EAAE,IAAI;aACjB,CAAC,CAAC;YAEH,MAAM,EAAE,IAAI,EAAE,IAAI,EAAE,UAAU,EAAE,MAAM,EAAE,GAAG,QAA2D,CAAC;YAEvG,IAAI,IAAI,IAAI,MAAM,GAAG,GAAG,EAAE,CAAC;gBACzB,MAAM,QAAQ,GAAG,gBAAgB,CAAC,IAAI,EAAE,QAAQ,CAAC,CAAC;gBAElD,uEAAuE;gBACvE,IAAI,QAAQ,CAAC,IAAI,CAAC,MAAM,GAAG,GAAG,EAAE,CAAC;oBAC/B,MAAM,YAAY,GAAG,MAAM,aAAa,CAAC,QAAQ,EAAE,QAAQ,EAAE,MAAM,EAAE,IAAI,CAAC,MAAM,CAAC,CAAC;oBAClF,IAAI,YAAY;wBAAE,OAAO,YAAY,CAAC,CAAC,0BAA0B;gBACnE,CAAC;YACH,CAAC;QACH,CAAC;QAAC,MAAM,CAAC;YACP,kEAAkE;QACpE,CAAC;QAED,yEAAyE;QACzE,MAAM,UAAU,GAAG,IAAI,CAAC,eAAe,EAAE,CAAC;QAC1C,MAAM,OAAO,GAAG,MAAM,IAAI,CAAC,aAAa,EAAE,CAAC;QAC3C,MAAM,WAAW,GAAG,IAAI,CAAC,yBAAyB,CAAC,UAAU,CAAC,CAAC;QAE/D,IAAI,IAAI,GAAgB,IAAI,CAAC;QAE7B,IAAI,CAAC;YACH,MAAM,OAAO,GAAG,MAAM,OAAO,CAAC,UAAU,CAAC;gBACvC,QAAQ,EAAE,EAAE,KAAK,EAAE,IAAI,EAAE,MAAM,EAAE,GAAG,EAAE;gBACtC,MAAM,EAAE,OAAO;gBACf,UAAU,EAAE,mBAAmB;aAChC,CAAC,CAAC;YACH,OAAO,CAAC,aAAa,CAAC,GAAG,EAAE;gBACzB,MAAM,CAAC,cAAc,CAAC,SAAS,EAAE,WAAW,EAAE,EAAE,GAAG,EAAE,GAAG,EAAE,CAAC,KAAK,EAAE,YAAY,EAAE,IAAI,EAAE,CAAC,CAAC;gBACvF,MAA0C,CAAC,MAAM,GAAG,EAAE,OAAO,EAAE,EAAE,EAAE,CAAC;YACvE,CAAC,CAAC,CAAC;YACH,IAAI,GAAG,MAAM,OAAO,CAAC,OAAO,EAAE,CAAC;YAC/B,MAAM,WAAW,GAAG,IAAI,CAAC,MAAM,CAAC,aAAa,IAAI,MAAM,CAAC;YACxD,MAAM,IAAI,CAAC,IAAI,CAAC,QAAQ,EAAE,EAAE,SAAS,EAAE,kBAAkB,EAAE,OAAO,EAAE,WAAW,EAAE,CAAC,CAAC;YAEnF,OAAO,MAAM,YAAY,CAAC,IAAI,EAAE,QAAQ,EAAE,MAAM,EAAE,IAAI,CAAC,MAAM,EAAE,WAAW,CAAC,CAAC;QAC9E,CAAC;QAAC,OAAO,GAAG,EAAE,CAAC;YACb,MAAM,QAAQ,GAAG,GAAG,YAAY,KAAK,CAAC,CAAC,CAAC,GAAG,CAAC,OAAO,CAAC,CAAC,CAAC,MAAM,CAAC,GAAG,CAAC,CAAC;YAClE,MAAM,MAAM,GAAsC;gBAChD,MAAM,EAAE,OAAO;gBACf,IAAI,EAAE,YAAY;gBAClB,IAAI,EAAE,IAAI;gBACV,MAAM,EAAE,EAAE;gBACV,UAAU,EAAE,CAAC;gBACb,OAAO,EAAE,EAAE;gBACX,KAAK,EAAE,EAAE,YAAY,EAAE,CAAC,EAAE,gBAAgB,EAAE,CAAC,EAAE,WAAW,EAAE,CAAC,EAAE,KAAK,EAAE,CAAC,EAAE;gBACzE,MAAM,EAAE,EAAE,gBAAgB,EAAE,CAAC,EAAE,cAAc,EAAE,CAAC,EAAE;gBAClD,KAAK,EAAE,QAAQ;aAChB,CAAC;YACF,MAAM,CAAC,MAAM,GAAG,cAAc,CAAC,MAAM,EAAE,QAAQ,EAAE,MAAM,CAAC,CAAC;YACzD,OAAO,MAAM,CAAC;QAChB,CAAC;gBAAS,CAAC;YACT,IAAI,IAAI,EAAE,CAAC;gBACT,MAAM,OAAO,GAAG,IAAI,CAAC,OAAO,EAAE,CAAC;gBAC/B,MAAM,IAAI,CAAC,KAAK,EAAE,CAAC,KAAK,CAAC,GAAG,EAAE,GAAE,CAAC,CAAC,CAAC;gBACnC,MAAM,OAAO,CAAC,KAAK,EAAE,CAAC,KAAK,CAAC,GAAG,EAAE,GAAE,CAAC,CAAC,CAAC;YACxC,CAAC;QACH,CAAC;IACH,CAAC;IAED,KAAK,CAAC,KAAK;QACT,IAAI,IAAI,CAAC,OAAO,EAAE,CAAC;YACjB,MAAM,IAAI,CAAC,OAAO,CAAC,KAAK,EAAE,CAAC,KAAK,CAAC,GAAG,EAAE,GAAE,CAAC,CAAC,CAAC;YAC3C,IAAI,CAAC,OAAO,GAAG,IAAI,CAAC;QACtB,CAAC;IACH,CAAC;CACF"}

package/dist/agent/loop.d.ts

@@ -1,6 +0,0 @@
-import type { Page } from "playwright-core";
-import type { AgentResult, PageSnapshot } from "../types.js";
-import type { ValidatedConfig } from "../config/schema.js";
-export declare function runStaticLoop(snapshot: PageSnapshot, url: string, prompt: string, config: ValidatedConfig): Promise<AgentResult | null>;
-export declare function runAgentLoop(page: Page, url: string, prompt: string, config: ValidatedConfig, getMemoryKb?: () => number): Promise<AgentResult>;
-//# sourceMappingURL=loop.d.ts.map

package/dist/agent/loop.d.ts.map

@@ -1 +0,0 @@
-{"version":3,"file":"loop.d.ts","sourceRoot":"","sources":["../../src/agent/loop.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,IAAI,EAAE,MAAM,iBAAiB,CAAC;AAC5C,OAAO,KAAK,EAAE,WAAW,EAAkD,YAAY,EAAE,MAAM,aAAa,CAAC;AAC7G,OAAO,KAAK,EAAE,eAAe,EAAE,MAAM,qBAAqB,CAAC;AAwD3D,wBAAsB,aAAa,CACjC,QAAQ,EAAE,YAAY,EACtB,GAAG,EAAE,MAAM,EACX,MAAM,EAAE,MAAM,EACd,MAAM,EAAE,eAAe,GACtB,OAAO,CAAC,WAAW,GAAG,IAAI,CAAC,CAsD7B;AASD,wBAAsB,YAAY,CAChC,IAAI,EAAE,IAAI,EACV,GAAG,EAAE,MAAM,EACX,MAAM,EAAE,MAAM,EACd,MAAM,EAAE,eAAe,EACvB,WAAW,GAAE,MAAM,MAAgB,GAClC,OAAO,CAAC,WAAW,CAAC,CAuGtB"}