abcmemory-vocabulary.mdc

---
description: User vocabulary abcmemory and abcremember for Cursor guidance files
alwaysApply: true
---

# abcmemory / abcremember vocabulary

## abcmemory

**abcmemory** means the committed files Cursor uses for agent guidance in this repo — **only** under `.cursor/` and the repo root files Cursor reads:

- `.cursor/skills/**`
- `.cursor/rules/**`
- `.cursor/prompts/**`
- `.cursor/hooks/**` and `.cursor/hooks.json` (when present)
- `.cursorrules`
- `.cursorignore`

## Not abcmemory

**`.llm/`** (plans, history, templates, context) is a **planning workspace**, not agent memory. Do not treat it as a second abcmemory tree.

Also not abcmemory unless you explicitly ask: `docs/`, `.github/`, and ad-hoc paths.

## abcremember

When the user says **abcremember**, **create or update** the best-fit abcmemory file(s) under **`.cursor/`** (or `.cursorrules` / `.cursorignore`): prefer updating existing guidance when the instruction extends the same topic; create a new skill, rule, or prompt only when no suitable file exists. Then follow that guidance in future work. Do **not** write to `.llm/` unless they explicitly say to remember something there.

For placement and workflow details, see `.cursor/skills/abcmemory/SKILL.md`.