Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add cafe3310/agent-skill-memories-off --skill memories-off-skillgit clone --depth 1 https://github.com/cafe3310/agent-skill-memories-offWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/cafe3310/agent-skill-memories-off/memories-off-skill)<a href="https://agentmods.dev/skills/cafe3310/agent-skill-memories-off/memories-off-skill"><img src="https://agentmods.dev/badge/skills/cafe3310/agent-skill-memories-off/memories-off-skill/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/cafe3310/agent-skill-memories-off/memories-off-skill"><img src="https://agentmods.dev/badge/skills/cafe3310/agent-skill-memories-off/memories-off-skill.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00047 | $0.01445 |
| Opus 5 | $0.00023 | $0.00723 |
| Sonnet 5 | $0.00009 | $0.00289 |
| Haiku 4.5 | $0.00005 | $0.00145 |
Grade A, and why
memories-off scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 11d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 91 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Agent Skill: memories-off (技能说明)
1. 概述 (Overview)
memories-off 是一个专门用于管理本地结构化知识库的 Agent Skill。它允许 LLM 将非结构化的对话、笔记、项目文档等,转化为由实体、关系和观察组成的知识图谱,并提供高效的检索和编辑能力。
该 Skill 旨在支持 LLM 实现“长期记忆”和“持续学习”,并基于 Markdown 文件提供本地优先的数据存储方案。
2. Agent 工作流与命令探索 (Agent Workflow)
作为 Agent,你不需要去阅读冗长的底层工具实现细节或规格文档。本 Skill 提供了自包含的命令行包装器 memocli。
2.1 任务初始化 (Explore First)
在处理任何与本知识库相关的任务前,必须首先运行以下命令获取全局上下文:
memocli explore
该命令会返回结构化的 XML 报告,包括:
- 知识库的全局描述和手册 (
meta.md)。 - 架构实体预览。
- 当前所有可用的
memocli子命令列表及简要说明。
2.2 了解命令用法 (Help System)
当你需要使用某个具体的子命令(如 search-entities, append-update)但不确定其参数时,请直接在终端运行该命令的帮助:
memocli <subcommand> --help
终端输出会告诉你最新的参数、选项以及调用示例,请严格按照 --help 提示的格式调用。
2.3 全局配置与多仓库管理 (Global Config & List Libraries)
如果您在多知识库场景下工作,可以通过在全局配置文件 ~/.config/memocli/config.yaml 中配置 libraries 来为各个库定义别名与描述。
配置文件示例如下:
libraries:
test_work:
path: /absolute/path/to/mock_kb_work
desc: 这是工作相关的知识库,包含任务跟踪和项目设计。
test_life:
path: /absolute/path/to/mock_kb_life
desc: 这是生活相关的知识库,包含日常琐事和健康记录。
配置完成后,您可以在执行任何子命令时使用 -l <alias>(或 --library <alias>)代替绝对路径,系统会自动进行别名展开。
此外,您也可以运行以下命令一键查询全局配置的所有知识库:
memocli list-libraries
3. Agent 最佳实践与约束 (Best Practices)
在使用 memocli 时,请严格遵守以下实践,以确保图谱的健康和数据的安全:
- 缓冲编辑优先: 在编辑已有实体内容时,严禁使用 Agent 自带的行定位编辑工具、也尤其禁止使用整体文件 Write。必须优先通过
memocli append-update以追加更新块的形式进行非破坏性修改。 - 多行写入使用 STDIN 模式: 当你需要通过
create-entity或append-update写入包含多行、引号、特殊符号、WikiLinks 等复杂内容时,不建议使用-c "..."传参,这极易导致 Shell 转义失败。建议使用管道(STDIN)模式。例如:echo "## 章节标题 - 内容1 - 内容2 [[链接]]" | memocli append-update --entity "实体名" --content-stdin --reason "更新理由" - 模糊检索优先 (Content-Grep First): 了解细节信息时,应使用
memocli search-entities。建议配合--type(过滤类型) 和--rel(过滤关系) 提升检索精度。 - 标题层级规范: 实体正文必须遵循严格的标题层级:文件顶部第一行固定为 H1 (
# 实体名),正文内部所有的业务章节标题必须统一标准化为 H2 (## 章节名)。禁止使用 H3 及更深层级。 - 优先用 Plain Text: 实体正文应当简洁。避免使用 Markdown 加粗 (
**)、斜体 (*) 等修饰性格式。允许使用 H2 标题、无序列表 (-) 和 WikiLinks ([[ ]])。 - 强制审计 (Audit): 所有修改操作必须提供简短的
--reason参数以供 Git 记录。 - 本地优先: 所有操作均基于本地文件系统,且当前路径若为知识库根目录,
memocli会自动推断--path .。
What ships with it
53 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
- assets/graph_template.html 13 KB
- assets/template-library/[范例.git]/说明.md 881 B
- assets/template-library/entities/2026-03-28-五一出行计划.md 703 B
- assets/template-library/entities/cafe3310.md 839 B
- assets/template-library/entities/Memory-Skill.md 768 B
- assets/template-library/entities/咩咩.md 290 B
- assets/template-library/entities/啾啾.md 275 B
- assets/template-library/entities/家庭大扫除.md 875 B
- assets/template-library/entities/渴望猫粮.md 632 B
- assets/template-library/entities/统一结果协议.md 235 B
- assets/template-library/meta.md 1.7 KB
- design-doc/关联定义/显式关系规范.md 1.8 KB
- design-doc/关联定义/隐式链接规范.md 2.1 KB
- design-doc/审计与日志定义/审计日志规范.md 1.6 KB
- design-doc/操作定义/图谱可视化规格.md 2.3 KB
- design-doc/操作定义/检索工具规格.md 2.6 KB
- design-doc/操作定义/管理工具规格.md 3.1 KB
- design-doc/操作定义/编辑工具规格.md 2.6 KB
- design-doc/操作定义/观察工具规格.md 2.2 KB
- design-doc/文档结构定义/元数据规范.md 1.9 KB
- design-doc/文档结构定义/实体结构规范.md 2.4 KB
- design-doc/文档结构定义/章节逻辑规范.md 2.0 KB
- design-doc/知识库结构定义/仓库布局规范.md 1.8 KB
- design-doc/知识库结构定义/手册结构规范.md 1.8 KB
- references/batch_operations.md 3.4 KB
- references/core_concepts.md 4.3 KB
- references/glossary.md 1.5 KB
- scripts/commands/__init__.py 0 B runs code
- scripts/commands/append_update.py 5.9 KB runs code
- scripts/commands/commit.py 1.9 KB runs code
- scripts/commands/create_entity.py 6.2 KB runs code
- scripts/commands/doctor.py 18 KB runs code
- scripts/commands/explore.py 8.5 KB runs code
- scripts/commands/find_doc_by_name.py 2.5 KB runs code
- scripts/commands/get_relations.py 4.8 KB runs code
- scripts/commands/graph.py 5.4 KB runs code
- scripts/commands/how_to_build_skill_upon.py 6.1 KB runs code
- scripts/commands/init.py 3.3 KB runs code
- scripts/commands/list_libraries.py 2.2 KB runs code
- scripts/commands/load_entities.py 3.6 KB runs code
- scripts/commands/manage_relations.py 4.6 KB runs code
- scripts/commands/merge_entities.py 7.8 KB runs code
- scripts/commands/merge_types.py 4.2 KB runs code
- scripts/commands/read_doc_by_name.py 2.9 KB runs code
- scripts/commands/rename_entity.py 4.9 KB runs code
- scripts/commands/rename_type.py 4.8 KB runs code
- scripts/commands/search_entities.py 8.3 KB runs code
- scripts/commands/squash_updates.py 3.4 KB runs code
- scripts/commands/stats.py 2.4 KB runs code
- scripts/install.py 10 KB runs code
- scripts/utility/__init__.py 0 B runs code
- scripts/utility/runtime.py 8.8 KB runs code
- scripts/utility/schema_define.py 12 KB runs code
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 11d ago First seen · 91 lines · 47 tokens per session scan A c2c5f39cf699
memories-off is a skill published in the GitHub repository cafe3310/agent-skill-memories-off (44 stars, last pushed 3mo ago), licensed Apache-2.0. It adds 47 tokens to every session and 1,445 once invoked, about $0.0002 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other skills, from other repositories
memoire-design-tooling
Use when a task spans interface understanding, design-system memory, UI audits, design CI, Figma, shadcn or Tailwind code generation, research, or agent design workflows and needs the correct Memi capability selected.
mama-context
Always-on background context injection from MAMA memory. Automatically surfaces relevant decisions when you work on code, without explicit invocation.
any-mcp
Use when an agent needs to read, search, create, edit, organize, upload, or chat with Anytype through the bounded any-mcp server. Covers safe tool selection, exact identifier chaining, optional toolsets, common PKM workflows, and the narrow anyr fallbacks for chat subscriptions and rich chat blocks.
mama-install
Install MAMA OS and follow its self-teaching CLI until the first report is confirmed.
metagit-context-pack
Token-efficient workspace onboarding via tiered context packs — map, repo cards, session digest, objectives, approvals, repomix profiles. Use at session start or when scoping work in a metagit-managed workspace.
codemem
Quick reference for all 32 codemem MCP tools. Use when working with the codemem knowledge graph — finding code, traversing relationships, storing memories, or running analysis.