[pull] master from botpress:master

integrations/browser/integration.definition.ts

@@ -5,7 +5,7 @@ import { actionDefinitions } from 'src/definitions/actions'
 export default new IntegrationDefinition({
   name: 'browser',
   title: 'Browser',
-  version: '0.7.0',
+  version: '0.8.0',
   description:
     'Capture screenshots and retrieve web page content with metadata for automated browsing and data extraction.',
   readme: 'hub.md',

integrations/browser/package.json

@@ -8,6 +8,7 @@
   "private": true,
   "dependencies": {
     "@botpress/sdk": "workspace:*",
+    "@mendable/firecrawl-js": "^4.3.8",
     "axios": "^1.7.2"
   },
   "devDependencies": {

integrations/browser/src/actions/browse-pages.ts

@@ -1,24 +1,8 @@
-import { IntegrationLogger } from '@botpress/sdk'
-import axios from 'axios'
+import { IntegrationLogger, RuntimeError } from '@botpress/sdk'
+import Firecrawl from '@mendable/firecrawl-js'
 import { FullPage } from 'src/definitions/actions'
 import * as bp from '.botpress'
 
-type FireCrawlResponse = {
-  success: boolean
-  data: {
-    markdown: string
-    metadata: {
-      ogLocaleAlternate: string[]
-      favicon?: string
-      title?: string | string[] | null
-      description?: string | string[] | null
-      sourceURL: string
-      error: string | null
-      statusCode: number
-    }
-  }
-}
-
 const COST_PER_PAGE = 0.0015
 
 const fixOutput = (val: unknown): string => {
@@ -37,37 +21,35 @@ const getPageContent = async (props: {
   timeout?: number
   maxAge?: number
 }): Promise<FullPage> => {
+  const firecrawl = new Firecrawl({ apiKey: bp.secrets.FIRECRAWL_API_KEY })
+
   const startTime = Date.now()
-  const { data: result } = await axios.post<FireCrawlResponse>(
-    'https://api.firecrawl.dev/v1/scrape',
-    {
-      url: props.url,
+
+  try {
+    const result = await firecrawl.scrape(props.url, {
+      fastMode: true,
       onlyMainContent: true,
+      maxAge: 60 * 60 * 24 * 7, // 1 week
+      removeBase64Images: true,
       waitFor: props.waitFor,
       timeout: props.timeout,
-      maxAge: 60 * 60 * 24 * 7, // 1 week
-    },
-    {
-      headers: {
-        Authorization: `Bearer ${bp.secrets.FIRECRAWL_API_KEY}`,
-      },
-    }
-  )
-
-  const { metadata, markdown } = result.data
+      formats: ['markdown', 'rawHtml'],
+      storeInCache: true,
+    })
 
-  props.logger.forBot().info(`Browsing ${props.url} took ${Date.now() - startTime}ms`, {
-    size: markdown.length,
-    statusCode: metadata.statusCode,
-    error: metadata.error,
-  })
+    props.logger.forBot().debug(`Firecrawl API call took ${Date.now() - startTime}ms for url: ${props.url}`)
 
-  return {
-    url: props.url,
-    content: markdown,
-    favicon: fixOutput(metadata.favicon),
-    title: fixOutput(metadata.title),
-    description: fixOutput(metadata.description),
+    return {
+      url: props.url,
+      content: result.markdown!,
+      raw: result.rawHtml!,
+      favicon: fixOutput(result.metadata?.favicon),
+      title: fixOutput(result.metadata?.title),
+      description: fixOutput(result.metadata?.description),
+    }
+  } catch (err) {
+    props.logger.error('There was an error while calling Firecrawl API.', err)
+    throw new RuntimeError(`There was an error while browsing the page: ${props.url}`)
   }
 }
 

integrations/browser/src/actions/discover-urls.ts

@@ -1,35 +1,13 @@
 import { RuntimeError } from '@botpress/client'
 import { IntegrationLogger, z, ZodIssueCode } from '@botpress/sdk'
-import axios from 'axios'
+import Firecrawl from '@mendable/firecrawl-js'
 import { isValidGlob, matchGlob } from '../utils/globs'
 import * as bp from '.botpress'
 
 const LAMBDA_TIMEOUT = 55_000
 
 const COST_PER_FIRECRAWL_MAP = 0.001
 
-type FirecrawlMapInput = {
-  /** The base URL to start crawling from */
-  url: string
-  /** Search query to use for mapping. During the Alpha phase, the 'smart' part of the search functionality is limited to 1000 search results. However, if map finds more results, there is no limit applied. */
-  search?: string
-  /** Ignore the website sitemap when crawling */
-  ignoreSitemap?: boolean
-  /** Only return links found in the website sitemap */
-  sitemapOnly?: boolean
-  /** In milliseconds */
-  timeout?: number
-  /** Max 30_000 */
-  limit?: number
-  /** Defaults to true */
-  includeSubdomains?: boolean
-}
-
-type FireCrawlResponse = {
-  success: boolean
-  links: string[]
-}
-
 type StopReason = Awaited<ReturnType<bp.IntegrationProps['actions']['discoverUrls']>>['stopReason']
 
 export const urlSchema = z.string().transform((url, ctx) => {
@@ -149,25 +127,16 @@ class Accumulator {
 }
 
 const firecrawlMap = async (props: { url: string; logger: IntegrationLogger; timeout: number }): Promise<string[]> => {
-  const { data: result } = await axios.post<FireCrawlResponse>(
-    'https://api.firecrawl.dev/v1/map',
-    {
-      url: props.url,
-      ignoreSitemap: false,
-      includeSubdomains: true,
-      sitemapOnly: false,
-      limit: 10_000,
-      timeout: Math.max(1000, props.timeout - 2000),
-    } satisfies FirecrawlMapInput,
-    {
-      signal: AbortSignal.timeout(Math.max(1000, props.timeout - 1000)),
-      headers: {
-        Authorization: `Bearer ${bp.secrets.FIRECRAWL_API_KEY}`,
-      },
-    }
-  )
+  const firecrawl = new Firecrawl({ apiKey: bp.secrets.FIRECRAWL_API_KEY })
+
+  const result = await firecrawl.map(props.url, {
+    sitemap: 'include',
+    limit: 10_000,
+    timeout: Math.max(1000, props.timeout - 2000),
+    includeSubdomains: true,
+  })
 
-  return result.links
+  return result.links.map((x) => x.url)
 }
 
 export const discoverUrls: bp.IntegrationProps['actions']['discoverUrls'] = async ({ input, logger, metadata }) => {

integrations/browser/src/definitions/actions.ts

@@ -30,6 +30,7 @@ const captureScreenshot: ActionDefinition = {
 const fullPage = z.object({
   url: z.string(),
   content: z.string(),
+  raw: z.string(),
   favicon: z.string().optional(),
   title: z.string().optional(),
   description: z.string().optional(),

packages/cognitive/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@botpress/cognitive",
-  "version": "0.1.44",
+  "version": "0.1.45",
   "description": "Wrapper around the Botpress Client to call LLMs",
   "main": "./dist/index.cjs",
   "module": "./dist/index.mjs",

packages/cognitive/src/client.ts

@@ -40,6 +40,7 @@ export class Cognitive {
   protected _provider: ModelProvider
   protected _downtimes: ModelPreferences['downtimes'] = []
   protected _useBeta: boolean = false
+  protected _debug = false
 
   private _events = createNanoEvents<Events>()
 
@@ -61,11 +62,14 @@ export class Cognitive {
       provider: this._provider,
       timeout: this._timeoutMs,
       maxRetries: this._maxRetries,
+      __debug: this._debug,
+      __experimental_beta: this._useBeta,
     })
 
     copy._models = [...this._models]
     copy._preferences = this._preferences ? { ...this._preferences } : null
     copy._downtimes = [...this._downtimes]
+
     copy.interceptors.request = this.interceptors.request
     copy.interceptors.response = this.interceptors.response
 

packages/cognitive/src/cognitive-v2/index.ts

@@ -11,6 +11,7 @@ type ClientProps = {
   botId?: string
   token?: string
   withCredentials?: boolean
+  debug?: boolean
   headers?: Record<string, string>
 }
 
@@ -21,12 +22,15 @@ type RequestOptions = {
 
 const isBrowser = () => typeof window !== 'undefined' && typeof window.fetch === 'function'
 
+export { Models } from './types'
+
 export class CognitiveBeta {
   private _axiosClient: AxiosInstance
   private readonly _apiUrl: string
   private readonly _timeout: number
   private readonly _withCredentials: boolean
   private readonly _headers: Record<string, string>
+  private readonly _debug: boolean = false
 
   public constructor(props: ClientProps) {
     this._apiUrl = props.apiUrl || 'https://api.botpress.cloud'
@@ -42,13 +46,28 @@ export class CognitiveBeta {
       this._headers['Authorization'] = `Bearer ${props.token}`
     }
 
+    if (props.debug) {
+      this._debug = true
+      this._headers['X-Debug'] = '1'
+    }
+
     this._axiosClient = axios.create({
       headers: this._headers,
       withCredentials: this._withCredentials,
       baseURL: this._apiUrl,
     })
   }
 
+  public clone(): CognitiveBeta {
+    return new CognitiveBeta({
+      apiUrl: this._apiUrl,
+      timeout: this._timeout,
+      withCredentials: this._withCredentials,
+      headers: this._headers,
+      debug: this._debug,
+    })
+  }
+
   public async generateText(input: CognitiveRequest, options: RequestOptions = {}) {
     const signal = options.signal ?? AbortSignal.timeout(this._timeout)
 

packages/cognitive/src/cognitive-v2/types.ts

@@ -1,21 +1,7 @@
-type Models =
+export type Models =
   | 'auto'
   | 'best'
   | 'fast'
-  | 'reasoning'
-  | 'cheapest'
-  | 'balance'
-  | 'recommended'
-  | 'reasoning'
-  | 'general-purpose'
-  | 'low-cost'
-  | 'vision'
-  | 'coding'
-  | 'function-calling'
-  | 'agents'
-  | 'storytelling'
-  | 'preview'
-  | 'roleplay'
   | 'anthropic:claude-3-5-haiku-20241022'
   | 'anthropic:claude-3-5-sonnet-20240620'
   | 'anthropic:claude-3-5-sonnet-20241022'
@@ -70,6 +56,7 @@ type Models =
   | 'xai:grok-4-fast-reasoning'
   | 'xai:grok-code-fast-1'
   | ({} & string)
+
 export type CognitiveRequest = {
   /**
    * @minItems 1

packages/cognitive/src/types.ts

@@ -49,6 +49,7 @@ export type CognitiveProps = {
   maxRetries?: number
   /** Whether to use the beta client. Restricted to authorized users. */
   __experimental_beta?: boolean
+  __debug?: boolean
 }
 
 export type Events = {

packages/llmz/package.json

@@ -2,7 +2,7 @@
   "name": "llmz",
   "type": "module",
   "description": "LLMz – An LLM-native Typescript VM built on top of Zui",
-  "version": "0.0.25",
+  "version": "0.0.26",
   "types": "./dist/index.d.ts",
   "main": "./dist/index.cjs",
   "module": "./dist/index.js",
@@ -67,7 +67,7 @@
   },
   "peerDependencies": {
     "@botpress/client": "^1.25.0",
-    "@botpress/cognitive": "^0.1.44",
+    "@botpress/cognitive": "^0.1.45",
     "@bpinternal/thicktoken": "^1.0.5",
     "@bpinternal/zui": "^1.0.1"
   },

packages/llmz/src/context.ts

@@ -1,3 +1,4 @@
+import { Models } from '@botpress/cognitive'
 import { z } from '@bpinternal/zui'
 import { cloneDeep, isPlainObject } from 'lodash-es'
 import { ulid } from 'ulid'
@@ -16,16 +17,14 @@ import { Transcript, TranscriptArray } from './transcript.js'
 import { wrapContent } from './truncator.js'
 import { ObjectMutation, Serializable, Trace } from './types.js'
 
-export type Model = 'best' | 'fast' | `${string}:${string}` | (string & {})
-
 export type IterationParameters = {
   transcript: TranscriptArray
   tools: Tool[]
   objects: ObjectInstance[]
   exits: Exit[]
   instructions?: string
   components: Component[]
-  model: Model
+  model: Models | Models[]
   temperature: number
 }
 
@@ -350,7 +349,7 @@ export namespace Iteration {
     messages: LLMzPrompts.Message[]
     code?: string
     traces: Trace[]
-    model: Model
+    model: Models | Models[]
     temperature: number
     variables: Record<string, any>
     started_ts: number
@@ -584,7 +583,7 @@ export class Context implements Serializable<Context.JSON> {
   public objects?: ValueOrGetter<ObjectInstance[], Context>
   public tools?: ValueOrGetter<Tool[], Context>
   public exits?: ValueOrGetter<Exit[], Context>
-  public model?: ValueOrGetter<Model, Context>
+  public model?: ValueOrGetter<Models | Models[], Context>
   public temperature: ValueOrGetter<number, Context>
 
   public version: Prompt = DualModePrompt
@@ -910,7 +909,7 @@ export class Context implements Serializable<Context.JSON> {
     exits?: ValueOrGetter<Exit[], Context>
     loop?: number
     temperature?: ValueOrGetter<number, Context>
-    model?: ValueOrGetter<Model, Context>
+    model?: ValueOrGetter<Models | Models[], Context>
     metadata?: Record<string, any>
     snapshot?: Snapshot
     timeout?: number