End-to-end encryptedAPI-firstOpen source

Encrypted docs
for AI agents

JSON documents and spreadsheets for agents — all end-to-end encrypted, all through a REST API. Model tickets and any other workflow as JSON conventions inside the same document system. Humans read and edit the same content in the browser.

Read the docs
REST API
POST /api/documentsCreate an encrypted document with a self access grant
GET /api/documentsList all documents accessible by your identity
GET /api/documents/:idGet a single document with its access grant
POST /api/documents/:id/editsAppend an encrypted patch + update the checkpoint snapshot
GET /api/documents/:id/editsList the edit history for a document
POST /api/documents/:id/shareGrant another identity access to a document
GET /api/identities/:idLook up an identity's public keys by ID

Prefer not to build your own client? Install the agentdocs skill from the Tank registry with safescript scripts for every endpoint.

Your agents forget everything

Every session starts from zero. Context, decisions, research — gone. You re-explain the same things to every agent, every time. Local tools don't help when your agents run in the cloud.

Without agentdocs
  • Context lost every session
  • Copy-paste between tools and prompts
  • Local files that cloud agents can't reach
  • Unencrypted data scattered across services
With agentdocs
  • +One workspace every agent reads and writes to
  • +Hosted API — any agent, anywhere, zero sync
  • +End-to-end encrypted — the server sees only ciphertext
  • +Knowledge compounds with every task

Four primitives, one API

Everything an agent needs to store, organize, and share structured knowledge — encrypted before it leaves your process.

Documents

Long-form markdown documents with full edit history. Every revision is stored so you can diff, audit, and roll back. Rendered in-browser with formatting.

Linked knowledge

Reference other documents by ID inside encrypted content. Agents build an interconnected knowledge graph — titles can change freely without breaking links.

Spreadsheets

Structured tabular data, encrypted cell-by-cell. Agents create and update rows through the API. Humans view and edit in a familiar grid.

Contacts

Securely store and manage an address book for your agents. Contact data is treated as encrypted JSON snapshots just like anything else.

JSON conventions

Treat everything as encrypted JSON documents. Tickets, specs, plans, and logs are all conventions in the same storage model.

Linked knowledge graph

Reference other documents by ID from inside encrypted content. Build interconnected documentation that survives renames and refactors.

Webhook events

Subscribe to document edits and sharing events. HMAC-signed payloads keep your integrations in sync.

Unlimited identities

Create as many cryptographic identities as you need. Each gets its own key pair. Share content between agents and humans with ECDH key exchange.

The shift

“Using LLMs to build personal knowledge bases for various topics of research interest. A large fraction of my recent token throughput is going less into manipulating code, and more into manipulating knowledge.”

“I think there is room here for an incredible new product instead of a hacky collection of scripts.”

— Andrej Karpathy

agentdocs is that product — but hosted, encrypted, and API-first. No local scripts, no file syncing. Your agents write to it from anywhere. The server never sees plaintext.

How it works

01

Register an identity

Generate Ed25519 + X25519 keys. Register via the API. Each agent or human gets their own cryptographic identity.

02

Create encrypted documents

POST encrypted JSON snapshots through the REST API. Every document gets a stable ID the agent keeps for future edits and links.

03

Link and cross-reference

Embed document IDs inside encrypted content to cross-reference pages and decisions. Links keep working across renames.

04

Share across identities

Grant access by wrapping the content's AES key with ECDH. Agents and humans collaborate on the same encrypted workspace.

Knowledge compounds

Every agent run adds to the knowledge base. Research notes link to decision docs. JSON ticket docs reference architecture pages. The graph grows denser and more useful with every task — and it's always one API call away from any agent, on any platform, in any cloud.

Day 1

An agent registers, creates its first wiki page, logs a decision. A single document in the graph.

Week 2

Dozens of linked pages. JSON ticket docs track open questions. A spreadsheet aggregates metrics. Agents cross-reference each other's work.

Month 3

A dense knowledge graph that new agents can onboard from instantly. Institutional memory that survives any single session, model, or provider.

Ed25519 signing|X25519 ECDH|AES-256-GCM|HKDF-SHA256|Web Crypto API

Give your agents a workspace

JSON docs and spreadsheets — encrypted and API-first. Open source. Free forever.