Skip to Content
CLImemory

flexkit memory

The Chat Agent and the MCP server keep one memory document per person and project: a Markdown summary of what they learned about you and your work (exact type names, data quirks, how you like answers). They read it at the start of every session and rewrite it as they learn. flexkit memory shows that document, lets you replace or edit it, and clears it.

Next to your own memory, the agent reads the project’s company memories: facts several members’ sessions confirmed. They are read-only here.

Synopsis

flexkit memory <subcommand> [arguments] [options]

Without a subcommand, flexkit memory runs show.

memory show

Show your preferred name, when the memory last changed, its size, the document itself and the company memories.

flexkit memory show [--content] [--json] [--project <id>]

--content prints only your document on stdout, so you can save it to a file.

Examples

# Show your memory
flexkit memory show
 
# Save it to a file to edit it
flexkit memory show --content > memory.md
 
# Print the company memories
flexkit memory show --json | jq -r '.companyMemories[]'

memory set

Replace your memory document with a Markdown file, or with stdin when you pass -. The whole document is replaced. The document is limited to 20,000 characters.

flexkit memory set <file|-> [--name <name>] [--json] [--project <id>]

--name also sets how the agent addresses you. Without it, the stored name is kept.

Examples

# Replace the memory with a file
flexkit memory set memory.md
 
# Also set your preferred name
flexkit memory set memory.md --name Dani
 
# Append a note from a script
{ flexkit memory show --content; echo "- Prices are in EUR"; } | flexkit memory set -

memory edit

Open your memory document in your editor and save it when the editor closes. The CLI uses $VISUAL, then $EDITOR, then vi (notepad on Windows). Nothing is saved when the document didn’t change or the editor exits with an error. edit needs a terminal; use memory set in scripts.

flexkit memory edit [--name <name>] [--project <id>]

Examples

# Edit the memory
flexkit memory edit
 
# Edit it in VS Code
EDITOR="code --wait" flexkit memory edit

memory clear

Empty your memory document. Your preferred name and the company memories stay. The CLI asks for confirmation; --yes skips it, and is required without a terminal. The agent starts learning again from your next sessions, and its hourly review of your recent chats may add notes back.

flexkit memory clear [--yes] [--json] [--project <id>]

Examples

# Clear the memory
flexkit memory clear
 
# Clear it without the prompt
flexkit memory clear --yes

Options

OptionApplies toDescription
--contentshowPrint only your memory document on stdout.
--name <name>set, editHow the agent addresses you, up to 100 characters.
--yes, -yclearSkip the confirmation prompt.
--jsonall except editPrint JSON on stdout.
--project <ID>, -pallProject to use. Default: the linked project.

See also the global options.

Permissions

Your memory is personal. Every project member can read and change their own memory, never someone else’s, and it needs a signed-in session from flexkit login. Project API tokens can’t read or change it, not even their creator’s, and are refused with A user session is required.

Last updated on