How to Integrate Your AI Agent with agentdocs

2026-06-07

by Uri Walevski

Most AI agent setups suffer from a basic security flaw. If you want your agent to read your personal notes, tasks, or spreadsheets, you usually have to connect them to a cloud-based document store like Google Docs or Notion.

This introduces a massive barrier: setting up official developer identities with Google is incredibly difficult. You are forced to jump through bureaucratic hoops, pay verification fees, or upload corporate and government documents just to get your integration approved and avoid scary security warnings.

Even if you survive the verification nightmare, you are still storing your private files in plaintext on their servers and handing over a powerful, long-lived API key to your agent loop. If your agent gets prompt-injected or compromised, your entire workspace is instantly exposed.

To solve this, we integrated prompt2bot with agentdocs.

---

What is agentdocs?

agentdocs is an open-source, end-to-end encrypted (e2ee) document and spreadsheet platform built for humans and AI agents.

This is important because traditional document storage is a security liability. When you store documents, spreadsheets, or personal notes in Google Docs or Notion, the hosting company can read your files in plaintext, and you have to trust them with your data.

With end-to-end encryption, your files are fully locked before they ever leave your device. The hosting server stores only scrambled, unreadable text and cannot read your document titles, contents, or spreadsheets. Only you and your authorized AI agents can decrypt and read them. This gives you complete privacy and guarantees that your automated assistants can read and write to your workspace without exposing your data to leaks.

Here is how to set up this integration for your agent in five simple steps.

1. Create a Cryptographic Identity

Go to agentdocs-nine.vercel.app and sign in.

Because agentdocs is fully encrypted and private, it uses cryptographic identities rather than usernames or passwords. Click "Create Identity" and give it a name (like "My Dev Identity" or "Personal Assistant"). This instantly generates your secure keys directly in your browser.

2. Export Your Identity Key Bundle

Your identity credentials act as your secret access key. To give your agent permission to read and write documents on your behalf, you need to export this bundle.

Click on your identity dropdown in the top right corner of the dashboard, and click Export Identity. Copy the resulting secure key string. Your agent will use this key to securely access your documents and handle all decryption locally.

3. Give the Key to Your Bot

Head over to your prompt2bot dashboard.

You can configure this in two ways:

• Option A: Paste the exported key bundle directly into the conversational builder AI chat and ask it to save it as a secret named agentdocs-identity.
• Option B: Open the bot settings, go to the dashboard secrets area, add a new secret with the name agentdocs-identity, and paste your key bundle there.

By saving it as a secure secret, your credentials remain fully encrypted. They are only injected into the secure execution sandbox at runtime—the LLM never sees your raw private keys, keeping your identity perfectly safe.

4. Add the agentdocs Skill

Next, ask the conversational builder AI to equip your bot with the agentdocs skill.

Once added, your bot can run document operations securely together with safescript.cc. Because Safescript executes these lightweight cryptographic and API operations natively on the edge, your bot doesn't need to spin up a full cloud VM. This means instant, secure execution with near-zero compute cost, allowing your agent to read, edit, and search your encrypted documents on demand.

5. Collaborate in Real-Time

Because the platform is built for both agents and humans, you can open any document created by your agent in the web UI.

Since you both share access to the encryption keys, you can edit a spreadsheet or write a markdown document, and your agent can instantly pull the decrypted updates to continue its task.

With end-to-end encryption, you get the collaborative power of a shared workspace with the absolute security of local-first cryptography.

← All posts