AI Agent Use Cases: Coding, Browser, Voice, Ops
Where procedural memory pays off. Six kinds of agent, what each one re-derives today, and what a stored procedure removes from the next run.
memorable.shCrawled Nov 13, 2026
Where procedural memory pays off. Six kinds of agent, what each one re-derives today, and what a stored procedure removes from the next run.
How Memorable captures a trace, extracts a procedure and recalls it. Install, consent, storage, and the two calls an agent makes.
The extraction and recall API: endpoints, authentication, payload shape, rate limits and errors. Two calls, store and recall.
Every memorable-cli command: init, enable, install-hooks, recall, ingest, show, skills, status. Install with npm i -g memorable-cli.
Running Memorable on a gbrain backend: procedures stored as pages in the user’s own database, with the measured results from the gbrain integration.
The system integrates procedural memory hooks with coding environments like Claude Code, Codex, Cursor, OpenCode, Devin, and Antigravity, enabling trace storage and recall.
Three teams running Memorable in production. What was measured on each, with the sample size and the statistical test, including the results that did not replicate.
gstack is Garry Tan's skill suite for coding agents. It runs on Claude Code with gbrain as its memory. A skill is a general manual loaded whole into context on every prompt. /investigate is 15,593 tokens. Memorable learned the same job from one session and handed back a 293-token procedure.
OpenHome builds an open smart speaker platform. Each device exposes around 150 keyword-triggered abilities that route into one agent, the superagent. The hardware has little onboard storage.
Quartermaster (QM) is an agent harness from YC that runs Codex in a Docker sandbox. Memorable stores procedures in QM's own Postgres. Every measured run used a fresh copy of the fixture, consent flipped read-only, and pass decided by sh test.sh run by the harness.
How Memorable turns an agent run into a procedure and hands it back, with the numbers from a public retrieval benchmark, two coding fixtures and a browser benchmark where the page changes between visits.
Memorable records the tool calls an agent made to finish a task and gives that procedure back the next time a similar task arrives. Fewer turns, fewer tool calls, same or better pass rate. Free CLI, works with Claude Code, Codex, Cursor, OpenCode, Devin and Antigravity.
Memorable stores and retrieves procedural memory within a user’s database, ensuring conversations remain on their machine.

Memorable is a procedural-memory layer for AI agents that records tool calls and reuses recorded workflows to guide similar tasks.