using-memory
/using-memory
Use the local memxt memory to recall past decisions, code, and context, and to persist new ones. Invoke whenever the user refers to prior work ("what did we decide", "how did we", "last time", "remember when") or makes a decision worth keeping.
Assessment and security
- Files in this package
- 1
- Scripts
- 0
- HTML comments
- 0
- URLs in SKILL.md
- 0
No signals detected. Repository changed on 23.07.2026, without references to retired models or APIs.
MIT, open: use and modify while retaining author attribution.
What it does and when to use it
Use the local memxt memory to recall past decisions, code, and context, and to persist new ones. Invoke whenever the user refers to prior work ("what did we decide", "how did we", "last time", "remember when") or makes a decision worth keeping.
Triggers from the description
Original author description
Use the local memxt memory to recall past decisions, code, and context, and to persist new ones. Invoke whenever the user refers to prior work ("what did we decide", "how did we", "last time", "remember when") or makes a decision worth keeping.
How it works
- Progressive disclosure (save tokens)
- Other tools
- When to recall
- When to store
- Principles
SKILL.md preview
---
name: using-memory
description: Use the local memxt memory to recall past decisions, code, and context, and to persist new ones. Invoke whenever the user refers to prior work ("what did we decide", "how did we", "last time", "remember when") or makes a decision worth keeping.
---
# Using memxt memory
You have a persistent, **local** memory palace via the `memory` MCP server.
It survives across sessions. Nothing leaves this machine. No cloud LLM is used
to compress or store memory (unlike claude-mem-style pipelines).
## Progressive disclosure (save tokens)
Always prefer this 2-step flow:
1. **`memory_search`** — default `detail=index` returns compact hits
(`#id`, wing/room, source, ~100-char snippet). Cheap.
2. **`memory_get`** — pass promising `id` or `ids` to load **full verbatim**
bodies only for what you need.
Do **not** pass `detail=full` unless the query is tiny or you already know you
need every body (wastes context).
Example:
```
memory_search({ query: "cart limit", limit: 5 })
→ #42 decisions/… "Cart is capped at 37…"
memory_get({ ids: [42] })
→ full drawer text
```
## Other tools
- **`memory_wake_up`** — continuity brief (also auto-injected at SessionStart).
- **`memory_profile`** — stable project facts; no embedding model.
- **`memory_store`** — persist a decision / constraint / fact (verbatim).
Prefer `room: "decisions"` for architecture choices.
- **`memory_forget`** — delete by drawer id.
- **`memory_stats`** / **`memory_dream`** — health / consolidation.
## When to recall
Before answering "what / why / how did we …", "last time", "remember when",
"our convention for …", or anything that depends on history:
1. `memory_search` (index)
2. `memory_get` on relevant ids
3. Cite `#id` / source when you use a memory
## When to store
After a real decision ("let's use X because Y"), a discovered constraint, a
correction the user makes, or a fact you'd want next session, call
`memory_store`. One crisp memory per fact.
Good:
> "Cart is capped at 37 items because the Brightwell ERP rejects larger orders (0x5C error)."
Bad: "we talked about the cart."
**Hooks already autosave** session tails on Stop + PreCompact (local, verbatim).
Still call `memory_store` for **important decisions** so they land in the
profile, not only as a long episode.
## Principles
- Memory is **verbatim** — never invent from a fuzzy summary.
- **Local-first**: embeddings + search on-device; zero network at query time.
- When unsure, **search** — index is cheap; full get is on demand.
Versions
The database has no other skills named using-memory.
Database snapshot: 13.11.2026 · package version 096bc77ee70b (SHA-256 digest of the entire package)
Downloads in the last 30 days: 0 · ranking
Report a problem with this skill: harmful content, incorrect descriptions or an author’s removal request.
Similar skills
- 86memory-recall
Recall previous conversations, decisions, scan tasks, user preferences, and cross-session context. Use when the user refers to earlier work, says "之前", "刚才", "上次", "以前", "那个任务", "其他会话", or asks what was discussed before. - 90daftari-memory
Use the user's Daftari vault as long-term memory. Use when the user refers to earlier decisions, preferences, or project context; asks you to remember, forget, or correct something; or when a conversation reaches a conclusion worth keeping for future chats. - 90sugar-memory
Use Sugar's local persistent memory in Hermes. Activate at the start of non-trivial work to recall prior decisions, error patterns and guidelines, before making a decision past sessions may already have settled, and after finishing work when something durable was learned. Do NOT… - 95memory-search
Search conversation history and semantic memory to recall previous discussions, decisions, and context. Use when the user asks to "search memory", "what did we discuss", "remember when", "find previous conversation", "check history", or before starting work to recall prior decisions. - 90session-recall
Search the unified local history of past Claude Code, Codex, and Cursor sessions at the START of a task when the user references a prior bug, feature, decision, file, or piece of work. Use before assuming fresh context for prompts such as "remember when…", "we worked on…", "the… - 100recall
Session recall using qrec CLI. Invoke when the user asks about past conversations, past implementations, previous decisions, "last time we...", "how did we...", "what was the approach for...", or any question about prior work that may have been discussed in a Claude Code session.