Skip to main content

xtrace-memory

Use XTrace long-term memory. Call xmem_search to pull relevant facts and past decisions when you lack context, and call xmem_recall BEFORE a risky tool action (editing a file, running a command, deploying) to surface procedural directives — lessons and procedures past sessions recorded about the tool or files you're about to touch.

ソース情報

リポジトリ
XTraceAI/xtrace-cli
ソースの最終更新活動
2026年7月22日 23:53
検出された SKILL.md の言語
英語
スター
1
フォーク
0

インストール方法

デフォルトでは、最初にソースを確認する Prompt が選択されています。直接コマンドに切り替えるか、ローカルコピーをダウンロードすることもできます。

ソースファイルを確認

インストールを決める前に、SKILL.md と SkillsMP に表示されている付属ファイルをお読みください。

SKILL.md を表示中

SKILL.md
ソースの指示 · 読み取り専用プレビュー
name
xtrace-memory
description
Use XTrace long-term memory. Call xmem_search to pull relevant facts and past decisions when you lack context, and call xmem_recall BEFORE a risky tool action (editing a file, running a command, deploying) to surface procedural directives — lessons and procedures past sessions recorded about the tool or files you're about to touch.
version
0.1.0
metadata
{"hermes":{"tags":["memory","xtrace","procedural-memory"],"category":"memory"}}
# XTrace memory (xmem) You have two memory tools backed by XTrace: - **`xmem_search(query)`** — semantic search over long-term memory (facts, decisions, prior context). Call it when you need background you don't already have in the conversation. - **`xmem_recall(tool, file?, entities?, task?)`** — procedural-memory recall. It fires on a *symbol tripwire*: pass the `tool` you're about to run and the `file`/`entities` it touches, and it returns `lesson`/`procedure` directives that earlier sessions recorded about exactly those symbols — what failed and the path that worked. ## When to use `xmem_recall` Call it **before** a consequential tool action, not after: - editing or creating a file → `xmem_recall(tool="Edit", file="<path>", task="<goal>")` - running a command / script → `xmem_recall(tool="Shell", entities=["<cmd or symbol>"])` - deploying, migrating, or touching infra → recall on the service/resource name Treat the returned directives as **advisory guardrails**: they flag known pitfalls and preferred procedures. If a directive is relevant, follow it; if it looks stale, note that and proceed. ## When to use `xmem_search` When you need context the current conversation doesn't contain — a past decision, a preference, how something was set up. Use a short, direct query.
GitHubで見る