Removed openAI, added instructions, updated readme

Author: Michael Liebmann

.env.example

@@ -2,3 +2,8 @@
 /dist
 .env
 .vercel
+.cursorrules
+# Ignore all .DS_Store files
+.DS_Store
+**/.DS_Store
+

.gitignore

@@ -1,9 +1,14 @@
 # Notion
 
-This plugin enables interaction with Notion-based knowledge repositories,
-allowing users to query and retrieve answers from both public and private Notion pages.
-The plugin integrates with OpenAI to provide high-certainty answers based on the content
-available in Notion and suggests follow-ups in case of missing content.
+Connery plugin for retrieving content from Notion pages.
+
+The Plugin currently contains only one action: `Get Notion Page Content`.
+
+- Allows you to chat with the content of a Notion page. Can be used for FAQ, Knowledge Base, etc.
+- The action extracts all content elements including text, media, and toggles, and returns the page content as a single string. It does not extract content form inline DBs.
+- Allows you to provide additional instructions for the Connery assistant on how to handle the content.
+  - This can be useful if the table contains additional information that should not be used for answering the question.
+  - It can also be used to provide more context or output formatting instructions.
 
 ## Repository structure
 

README.md

@@ -1,18 +1,17 @@
 import { ActionDefinition, ActionContext, OutputObject } from 'connery';
 import { Client, iteratePaginatedAPI, isFullBlock } from '@notionhq/client'; // Import Client and types from Notion
-import OpenAI from 'openai';
 
 const actionDefinition: ActionDefinition = {
   key: 'askNotionPage',
-  name: 'Ask Notion Page',
+  name: 'Get Notion Page Content',
   description:
-    'This action enables users to ask questions and receive answers from a knowledge base hosted on a private Notion page. The action accesses the Notion page via its URL using the Notion API and an API key. Users’ questions are processed by OpenAI, which generates answers based on the content retrieved from the page. The action supports all content elements, including toggles.',
+    'This action retrieves the content of a Notion page using its URL and the Notion API. It can optionally include instructions before the page content. The action required the Notion page URL and Notion API key connected to this URL. It fetches all content elements including text, media, and toggles, and returns the page content as a single string. It does not extract content form inline DBs.',
   type: 'read',
   inputParameters: [
     {
       key: 'notionPageUrl',
       name: 'Notion Page URL',
-      description: 'The URL of the private Notion page to fetch knowledge content from.',
+      description: 'The URL of the private Notion page to fetch content from.',
       type: 'string',
       validation: {
         required: true,
@@ -28,30 +27,12 @@ const actionDefinition: ActionDefinition = {
       },
     },
     {
-      key: 'openaiApiKey',
-      name: 'OpenAI API Key',
-      description: 'API key to authenticate with OpenAI',
+      key: 'instructions',
+      name: 'Instructions',
+      description: 'Optional instructions for content processing.',
       type: 'string',
       validation: {
-        required: true,
-      },
-    },
-    {
-      key: 'openaiModel',
-      name: 'OpenAI Model',
-      description: 'The model to use for generating the answer (e.g. gpt-4-turbo, gpt-4o-mini, etc.).',
-      type: 'string',
-      validation: {
-        required: true,
-      },
-    },
-    {
-      key: 'question',
-      name: 'User Question',
-      description: 'The question asked by the user about the particular knowledge base.',
-      type: 'string',
-      validation: {
-        required: true,
+        required: false,
       },
     },
   ],
@@ -60,8 +41,8 @@ const actionDefinition: ActionDefinition = {
   },
   outputParameters: [
     {
-      key: 'textResponse',
-      name: 'Text response',
+      key: 'notionContent',
+      name: 'Notion Content',
       type: 'string',
       validation: {
         required: true,
@@ -93,73 +74,23 @@ export async function handler({ input }: ActionContext): Promise<OutputObject> {
       );
     }
 
-    // Ask OpenAI for an answer
-    const answer = await askOpenAI(input.openaiApiKey, input.openaiModel, pageContent, input.question);
+    // Prepare the output based on whether instructions are provided
+    let output: string;
+    if (input.instructions) {
+      output = `Follow these instructions: ${input.instructions}\nContent: ${pageContent}`;
+    } else {
+      output = pageContent;
+    }
 
-    // Return the model's answer directly
-    return { textResponse: answer };
+    // Return the formatted output
+    return { notionContent: output };
   } catch (error: any) {
     console.error('An error occurred:', (error as Error).message);
     throw new Error(`Error occurred: ${(error as Error).message}`);
   }
 }
 
-async function askOpenAI(
-  openaiApiKey: string,
-  openaiModel: string,
-  pageContent: string,
-  question: string,
-): Promise<string> {
-  // Initialize OpenAI with the provided API key
-  const openai = new OpenAI({ apiKey: openaiApiKey });
-
-  // Create the system message with instructions for the model
-  const systemMessage = `You are an FAQ expert. When asked a question or given a request related to a specific topic, you provide an accurate and concise answer based strictly on the content provided. 
-    You respond in the same language as the user’s input and adjust your answer to fit the context of the request, whether it’s a direct question or an indirect inquiry.
-    You never guess or paraphrase — only answer if the explicit content for that request is available. 
-    If there are any disclaimers or indications in the content that it should not be shared with clients or is a work in progress, include that information only if it is explicitly mentioned. 
-    Here is the content you should use to generate your answer:
-    ”${pageContent}”
-    `;
-
-  // Set the user's question separately
-  const userQuestion = `Based on this content, please respond to the following request or question with high confidence:
-    ”${question}”. 
-    If you are not confident that the content fully addresses the request, respond with: 
-    ‘I don’t have enough information to answer your question.’
-    `;
-
-  // Request completion from OpenAI using the specified model
-  const response = await openai.chat.completions.create({
-    model: openaiModel,
-    messages: [
-      { role: 'system', content: systemMessage },
-      { role: 'user', content: userQuestion },
-    ],
-  });
-
-  // Log and handle the response
-  if (!response.choices || response.choices.length === 0) {
-    console.error('Model did not respond with any choices.');
-    throw new Error('Model did not respond.');
-  }
-
-  const messageContent = response.choices[0].message.content;
-
-  if (messageContent === null || messageContent.trim().length === 0) {
-    console.error("Model's answer length is too short.");
-    throw new Error("Model's answer is too short.");
-  }
-
-  const answer = messageContent.trim();
-
-  return answer;
-}
-
-/**
- * Helper function to retrieve all blocks from a Notion page using pagination.
- * Recursively fetches child blocks if they exist.
- */
+// Helper function to retrieve all blocks from a Notion page using pagination. Recursively fetches child blocks if they exist.
 async function retrieveBlockChildren(notion: Client, id: string) {
   const blocks: Array<any> = [];
   for await (const block of iteratePaginatedAPI(notion.blocks.children.list, { block_id: id })) {
@@ -174,18 +105,12 @@ async function retrieveBlockChildren(notion: Client, id: string) {
   return blocks;
 }
 
-/**
- * Helper function to extract plain text from a rich text object in Notion.
- * Combines all pieces of text within a block into a single string.
- */
+// Helper function to extract plain text from a rich text object in Notion. Combines all pieces of text within a block into a single string.
 const getPlainTextFromRichText = (richText: any) => {
   return richText.map((t: any) => t.plain_text).join('');
 };
 
-/**
- * Helper function to convert a Notion block into a string representation.
- * Handles various block types, including media, tables, and text blocks.
- */
+// Helper function to convert a Notion block into a string representation. Handles various block types, including media, tables, and text blocks.
 const getTextFromBlock = (block: any) => {
   let text;
 
@@ -248,10 +173,7 @@ const getTextFromBlock = (block: any) => {
   return block.type + ': ' + text;
 };
 
-/**
- * Helper function to extract the source text of media blocks, such as images or videos,
- * including any associated captions.
- */
+//Helper function to extract the source text of media blocks, such as images or videos, including any associated captions.
 const getMediaSourceText = (block: any) => {
   let source, caption;
 
@@ -273,10 +195,7 @@ const getMediaSourceText = (block: any) => {
   return source;
 };
 
-/**
- * Helper function to extract the Notion page ID from the provided URL.
- * The function uses a regular expression to identify and return the page ID.
- */
+// Helper function to extract the Notion page ID from the provided URL. The function uses a regular expression to identify and return the page ID.
 function extractPageIdFromUrl(url: string): string {
   const regex = /([a-f0-9]{32})|([a-f0-9]{8}-[a-f0-9]{4}-[a-f0-9]{4}-[a-f0-9]{4}-[a-f0-9]{12})/;
   const match = url.match(regex);