Choose a deliberate write boundary
Grant read access, propose-only access, or direct writing. Proposed writes, appends, and restores wait for a person to accept them in Review.
Review and history
Let an agent help. Keep the review in your hands.
Some work benefits from an agent draft and a human decision. Choose propose-only access for suggested note changes, then inspect the proposal and the history behind the current note.
See who changed what
Version history records the actor, time, optional reason, and a diff. Activity also helps you find the note behind an agent action.
Recover an earlier version
Restore copies an old version forward as a new version. Earlier history remains available so recovery does not erase the record of changes.
FROM IDEA TO USEFUL CONTEXT
Put it into practice.
Set the access level
Choose Propose changes when connecting an agent or creating its token.
Inspect the suggestion
Open Review, check the proposed operation, reason, and current note, then accept or reject it.
Follow the history
Use the new version and its attribution to understand the result or restore an earlier version if needed.
A FEW USEFUL DETAILS
Questions, answered.
Can propose-only agents perform every write operation?
No. Propose-only access supports suggested note writes, appends, and restores. Operations such as creating Blocks, changing comments, or completing tasks need direct write access.
Does restoring erase later history?
No. Restoring copies the chosen version into a new current version and keeps the existing history.
KEEP EXPLORING
Connect the next piece.
Discussions and Inbox
Note discussions and an inbox for agent changes
Discuss saved notes in Docblock with threads and selected-text comments. Use Inbox to review workspace discussions and changes made by agents.
Explore Note discussions and an inbox for agent changesAI agent memory
Shared memory for AI agents over MCP
Give supported AI agents a shared Docblock memory over MCP, with searchable Markdown notes, daily appends, source retrieval, access scopes, and review.
Explore Shared memory for AI agents over MCPClaude Code integration
Connect Claude Code to shared project memory
Use Docblock as shared note context for Claude Code over remote MCP. Read project decisions, find references, and propose or write a session handoff.
Explore Connect Claude Code to shared project memoryA PLACE TO BUILD ON
Give your next thought a home.
Start with a note. Add structure, connect an agent, and keep the useful context together.