DreamLake

Create and find memos

Quick start

Log in with dl login. Core examples require CLI 0.29.0+ and a compatible server; individual guides state newer feature requirements. Check your installed version with dl --version.

Before an agent's first live read or edit, set its task identity. Set both DREAMLAKE_AGENT_ID and DREAMLAKE_AGENT_NAME and reuse them for the task.

bash
# Find a note and read it directly.
dl memos search "release plan"
dl memos read release-plan

# Preview a unique wording change, apply it, then read back.
dl memos replace "Draft checklist" --text "Ready for review" --note release-plan --dry-run
dl memos replace "Draft checklist" --text "Ready for review" --note release-plan
dl memos read release-plan

# Create a private note from a Markdown file.
dl memos create "Release plan" --file release-plan.md

# Cross-interface design (unreleased target APIs; verify installed --help):
# https://docs.dreamlake.ai/notes/api/

Read normal output directly; no --json or --view is needed. Retain the read's revision and hash for an edit. The replacement helper requires exactly one match and guards its own read/write window. A dry run does not reserve the memo. For an edit prepared from an older snapshot, use a patch with the original baseline revision.

bash
dl memos list --limit 10
dl memos list --shared
dl memos list --namespace acme
dl memos list --include-archived

--ns is an alias of --namespace on every Memos command; both are the source namespace to look in, never a destination. --offset N pages by rows (--limit 20 --offset 20 is the second page). The CLI looks a memo ID up in --ns or your configured namespace. Without --ns, read-only commands follow an ID to its owner when it is not found there (Reading); writes do not, so pass --ns to change another workspace's memo. The TypeScript facade (dl.notes(noteId)) discovers the owner itself.

bash
dl memos search "release plan"
dl memos search deploy --namespace acme --json
dl memos search deploy --include-archived

The default namespace is personal. Add --namespace acme for an organization's memos. --shared lists memos shared with you across namespaces; read them using the owner's namespace. Use IDs, slugs, or exact titles; prefer full IDs in scripts because titles can repeat. Slug/title lookup includes archived memos and follows catalog pages. Search matches titles and indexed bodies by case-insensitive substring.

Create or follow a memo

bash
dl memos create "Release plan" --text "Draft checklist"
printf '# Release plan\n' | dl memos create "Release plan" --file - --json

The title is optional: dl memos create --ns your-workspace creates an empty memo named Untitled; --file - reads the initial source from stdin.

New memos are private unless you pass --public. Repeated titles get distinct slugs; keep the returned ID. Use a file or printf for multiline text.

bash
dl memos read release-plan
# Inspect source with generated address hints; do not write the hints back.
dl memos read release-plan --view markdown
# Preserve exact source plus revision/hash for a programmatic edit baseline.
dl memos read release-plan --json > baseline.json
dl memos read release-plan --since "10 minutes ago" --format diff
# Detailed read design (unreleased target syntax):
# https://docs.dreamlake.ai/notes/api/reading/
# Reuse a stable agent identity for this task.
export DREAMLAKE_AGENT_ID=release-reviewer-a
export DREAMLAKE_AGENT_NAME="Release reviewer A"
dl memos read release-plan --linger
# Ctrl-C stops following.

Reads require read access; edits require write access. An inaccessible memo may report as not found. The CLI reads server content and cannot recover an unsynced browser draft. For older body/section scripts, see Legacy commands.

Current commands and the shared design

The cross-interface API design defines the proposed CLI, TypeScript, Python and tool contract. Its examples are explicitly unreleased. The commands on this page retain the current CLI syntax; for example, creation still takes a name and live following uses read --linger. The Memos index compares current behavior with the target before you adapt an existing script.