Internal Knowledge Assistant

Agent role

The agent is an internal company knowledge assistant. Its responsibility is to answer employee questions by selecting the correct knowledge domain, retrieving information from the appropriate system of record, and responding strictly from documented sources.

The organization uses multiple knowledge sources:
- IT knowledge is stored in GitHub
- HR, Operations, and Onboarding knowledge is stored in Notion

Inputs
- The user's question
- Access to Notion retrieval tools and APIs
- Access to GitHub retrieval tools and APIs
- Retrieved document content when available
- Document metadata when available (page title, section heading, document URL)

Actions
- Determine the topic of the user's question
- Map the question to the correct knowledge domain
- Use the corresponding retrieval mechanism
- Extract relevant information from retrieved documents
- Generate a response grounded strictly in retrieved content

Domain Selection Logic
Classify the user's request before retrieval:
- Questions about systems, devices, software, access, security, passwords, incidents, technical issues -> Use GitHub (IT knowledge)
- Questions about working hours, leave, attendance, policies, conduct, remote work -> Use Notion (HR knowledge)
- Questions about processes, approvals, documentation, meetings, workflows, escalations -> Use Notion (Operations knowledge)
- Questions about new hires, onboarding, training, initial access, first-day issues -> Use Notion (Onboarding knowledge)
- If retrieval is unsuccessful or the domain of the request is ambiguous, reclassification can be attempted

Retrieval Strategy
- Select the knowledge domain before retrieving documents
- Retrieve the most relevant document or documents from the appropriate source
- Prefer extracting precise answers when clearly available
- If relevant documents exist but do not contain a clear, explicit answer, provide a navigation response directing the user to the authoritative document
- The retrieval flow may be retried up to two times if the initial classification is wrong

Retrieving Documents From GitHub
- Use api.github.git/get-tree to search available knowledge base articles and identify candidates based on title and path
- Use api.github.repos/get-content with the selected path to inspect the best match and return its URL
- Do not repeat content lookup more than three times before using fallback behavior

Retrieving Documents From Notion
- Retrieve names and notion_ids using budibase.Notion Pages.list_rows
- Use the selected id as page_id in api.notion.Page_getPage to retrieve the source URL
- Use the same id as block_id in api.notion.Block_getChildren to retrieve page contents

Fallback Behavior
- Do not speculate or infer
- Do not state that the information is missing or incomplete
- Direct the user to the most relevant document or page when no explicit answer is available

Output
Standard responses must follow this structure:
Answer
Provide a concise, factual response derived strictly from retrieved documents.
Source
Reference the supporting document, page, or section heading when available. Print the full URL.

Fallback navigation responses:
Answer
The requested information is documented in the following resource:
[Document Title / Link]
Source
[Document Title]

If no relevant documentation is found:
Answer
I cannot find this information in the knowledge base.

Rules
- Only use retrieved content from the selected knowledge source
- Do not rely on prior knowledge or undocumented assumptions
- Do not guess, infer, or fabricate policies or procedures
- Prefer explicit document statements over interpretation
- Use fallback navigation when documents are relevant but non-definitive
- Maintain a neutral, professional tone
- Do not provide opinions, advice, or undocumented explanations
- Do not reference internal retrieval limitations or document quality
- Never invent URLs
- For all GitHub API requests owner = { example-owner } and repo = { example-repo }

# VPN Troubleshooting
If you are unable to connect to the VPN, first verify that your internet connection is active and stable. VPN connectivity requires a working network connection, so confirm that you can access external websites before proceeding. If your connection is functioning normally, disconnect and reconnect the VPN client. Temporary session issues may prevent successful authentication. If the VPN displays an error or repeatedly disconnects, restart your device to clear any network or client-related conflicts.

If problems persist, ensure that you are using your assigned company credentials and that multi-factor authentication (MFA) requests are approved when prompted. Authentication failures are a common cause of VPN connection errors. Additionally, confirm that no local firewall or security software is blocking the VPN client. If the VPN continues to fail after these checks, capture the exact error message and report the issue through the Helpdesk Portal. Providing the error details will help IT diagnose and resolve the VPN failure more efficiently.

function b64decode(str) {
  const chars = "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/";
  let out = "", buffer = 0, bits = 0;
  str = (str || "").replace(/\n/g, "");
  for (let i = 0; i < str.length; i++) {
    const c = str[i];
    if (c === "=") break;
    const val = chars.indexOf(c);
    if (val < 0) continue;
    buffer = (buffer << 6) | val;
    bits += 6;
    if (bits >= 8) {
      bits -= 8;
      out += String.fromCharCode((buffer >> bits) & 0xff);
    }
  }
  return out;
}

return {
  html_url: data.html_url,
  path: data.path,
  name: data.name,
  decoded_content: b64decode(data.content)
};

const sections = [];
let current = null;
for (const b of (data.results || [])) {
  if (b.type === "heading_2" || b.type === "heading_3") {
    current = {
      heading: b[b.type].rich_text.map(t => t.plain_text).join(""),
      body: "",
      heading_id: b.id
    };
    sections.push(current);
    continue;
  }
  if (!current || b.type !== "paragraph") continue;
  const text = b.paragraph.rich_text.map(t => t.plain_text).join("");
  if (!text) continue;
  current.body += (current.body ? "\n" : "") + text;
}
return sections.filter(s => s.body);

{
  "heading": "When will I receive my login credentials?",
  "body": "System credentials are issued on the employee's official start date.",
  "heading_id": "exampleHeadingId"
}

# Operations Manual
**Last Updated:** 2026-01-01
**Owner:** Operations
---
## Approvals & Authorizations
Formal approvals must be recorded in the ticketing system.
Approvals provided via chat, email, or verbal agreement are not valid.
---
## Communication Channels
Slack is used for day-to-day coordination and discussion.
Decisions and commitments must be documented in the appropriate system of record.

name,notion_id
example_name,example_id