Structured storage · MCP-native
Qendaro stores your docs, notes, and specs as structured items — and ships an MCP server, so Claude Code, Cursor, or any agent reads and writes them with your permissions. Every change is attributed and reversible.
Free while in beta · Sign up with email or Google
>Add yesterday's session timeout decision to the auth spec
search_items("session timeout")
get_item("Auth Spec — Sessions")
update_item("Auth Spec — Sessions", …)
✓ Spec updated — change logged to revision history
Idle sessions expire after 30 minutes. Refresh tokens rotate on use and are revoked on logout.
v12 · Claude Code · on behalf of Ivan · just now
How it works
Create a collection and define its layers and item types — or start from a preset like Specification, Test Cases, or Sources and adjust as you go.
Add one MCP endpoint to Claude Code, Cursor, or Claude Desktop. OAuth sign-in with PKCE — your agent authenticates as you, no API keys to paste around.
Edit in the app while agents read, write, and comment through MCP. Same permissions, one shared history, every change attributed to who — or what — made it.
One engine, your structure
A collection is whatever you need it to be. Layers group your item types; each item type carries its own field schema — markdown, embedded sub-forms, references, computed fields. Your agent receives the schema through MCP, so it writes structured items, not wall-of-text notes.
Features
Define your own layers, item types, and field schemas per collection. The same engine holds meeting notes, project docs, a lightweight CRM, or a full requirements tree — nothing is hardcoded.
The MCP server forwards your token on every call, so an agent holds exactly your permissions. A reader's agent can only read. No separate bot accounts, no over-scoped API keys.
Every change — by a person or an agent — lands in an append-only log with full attribution. Browse revisions, see word-level diffs, revert a single change without losing later edits.
Link items, fields, or exact text fragments across layers — spec to test case, decision to source. When an anchor's text changes, Qendaro flags the trace for review instead of silently breaking it.
Threaded comments anchored to items, fields, or selections. Mention @Agent to hand work to your AI, or @teammates to pull them in — everyone sees the same thread.
Google-Docs-style co-editing on markdown fields, live presence, and instant updates across tabs. Your edits and your agent's edits merge — no lock-outs, no lost work.
Connect in one command
Add Qendaro to your agent, sign in through OAuth in the browser, and it's connected — searching, creating, and updating items, managing structure, leaving comments, and linking traces. Agent activity shows up in the app labeled as the agent, acting on your behalf.
$ claude mcp add --transport http qendaro \
https://mcp.qendaro.com/mcp
# sign in once in the browser — done
Create a workspace, shape your first collection, and connect your agent — free while Qendaro is in beta.