cloudycotton e207d36afb prepare for skill publishing
- Move SKILL.md to skill/ for skills.sh convention
- Update package.json: @runablehq/mem, add description/keywords/files
- Add terminal demo gif showing skills add + mem usage
- Rewrite README with demo gif, skills.sh and runable.com links
- Add VHS tape file for regenerating the gif
2026-03-21 19:11:46 +05:30
2026-02-11 14:14:18 +05:30
2026-03-21 19:11:46 +05:30
2026-03-21 19:11:46 +05:30
2026-02-11 14:14:18 +05:30
2026-02-11 14:14:18 +05:30
2026-02-11 14:14:18 +05:30
2026-02-11 14:14:18 +05:30
2026-02-11 14:14:18 +05:30
2026-03-21 19:11:46 +05:30
2026-03-21 19:11:46 +05:30
2026-02-11 14:14:18 +05:30

mem · agent memory store

Three operators, one tool. Remember, recall, forget — with full-text search backed by SQLite.

mem demo — install skill, remember, recall, search

skills.sh runable

Setup

npx skills add runablehq/memory --all --global

This installs the mem skill for all your agents via skills.sh. Your agent will automatically know how to store and recall information across sessions.

Manual install

# requires bun
bun build --compile src/cli.ts --outfile dist/mem
cp dist/mem ~/.local/bin/mem

Usage

mem [query]                     recall (search, list, or get by ID)
mem + "content" [--tag a,b]     remember
mem - <id>                      forget

Remember

mem + "chose SQLite for simplicity" --tag architecture
mem + "user prefers dark mode" --tag prefs
mem + "deploy: bun build --compile" --tag deploy
mem + --image ./screenshot.png --title "Current UI" --tag ui
echo "long content" | mem + --tag notes

Recall

mem                             # list recent
mem "deploy"                    # search
mem 7sjtNVyZrNIa                # get by ID
mem --tag prefs                 # filter by tag
mem "api" --limit 5 --json     # limit + JSON output

Forget

mem - 7sjtNVyZrNIa              # forget one
mem - id1 id2 id3               # forget many

Piping

mem "old" --json | jq .         # pipe to jq
mem "old" --json | jq -r '.[].id' | xargs -I{} mem - {}
echo "long content" | mem + --tag notes

Flags

--tag a,b     filter (read) or apply (write)
--json        structured JSON output
--full        full content inline
--limit n     max results (default 20)
--image path  attach image
--title text  set title
--export dir  export image

Storage

~/.mem/mem.db — SQLite with FTS5 full-text search.

S
Description
Store and retrieve memories (notes, facts, decisions, snippets, images) using a local SQLite database with full-text search. Use when you need to remember…
Readme 856 KiB
Languages
TypeScript 100%