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 kangarooking/system-prompt-skills --skill memory-systemgit clone --depth 1 https://github.com/kangarooking/system-prompt-skillsWrote 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/kangarooking/system-prompt-skills/memory-system)<a href="https://agentmods.dev/skills/kangarooking/system-prompt-skills/memory-system"><img src="https://agentmods.dev/badge/skills/kangarooking/system-prompt-skills/memory-system.svg" alt="Measured on agentmods" 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 | $0.00166 | $0.02515 |
| Opus 5 | $0.00083 | $0.01257 |
| Sonnet 5 | $0.00033 | $0.00503 |
| Haiku 4.5 | $0.00017 | $0.00251 |
Grade A, and why
memory-system 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 8d 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 — 117 lines — stays where its author put it; the contents beside it link to each section on GitHub.
记忆与个性化架构 (Memory System)
R — 原文 (Reading)
Claude Web: userMemories 注入 + memory_user_edits tool + 选择性应用规则 + 静默归因 + 边界示例 Claude Code: File-based persistent memory with typed memories (user, feedback, project, reference) + MEMORY.md index Claude Opus 4.7: "NEVER reference sensitive memories in unrelated contexts" + bad example (proactively mentioning deceased pet) GPT-4o/GPT-4.5: bio tool for persistence + sensitive data prohibition FlintK12: Pedagogical memory (interests, preferences, grade level) + mandatory create_memory call Gemini CLI: save_memory tool + GEMINI.md files for project context
I — 方法论骨架 (Interpretation)
记忆系统的核心设计模式围绕"生命周期管理"和"边界控制"两个轴展开：
记忆生命周期 (CRUL 模型):
- Create (创建): 决定何时创建记忆。FlintK12 采用"强制创建"策略（mandatory create_memory call），确保关键教学信息不被遗漏。Claude Web 采用"选择性创建"——只在用户明确表达偏好时创建。
- Retrieve (检索): 在每轮对话开始时，将相关记忆注入 context window。关键设计决策是"注入多少"和"注入什么"——全部注入会消耗上下文窗口，选择性注入需要相关性判断。
- Apply (应用): 静默地应用记忆内容来调整回复，而不在回复中显式引用。Claude Opus 4.7 的反面教材极为重要：如果用户曾提到宠物去世，AI 不应在无关对话中主动提及。
- Update (更新): 允许用户编辑或删除已存储的记忆。Claude Web 的 memory_user_edits 工具和 Claude Code 的 typed memory 系统都支持这一能力。
类型化记忆 (Typed Memory): Claude Code 将记忆分为四类：user（用户偏好）、feedback（交互反馈）、project（项目上下文）、reference（参考文档）。不同类型有不同的存储策略、检索优先级和应用规则。
边界控制: 记忆系统最大的风险不是"记不住"，而是"在不该记住的时候记住了"或"在不该引用的时候引用了"。
A1 — 案例分析 (Past Application)
案例 1: Claude Web 的静默记忆应用
- 问题: 如何让 AI 利用用户记忆但不让对话变得尴尬或侵犯隐私？
- 设计模式的使用: Claude Web 实现了"静默归因"策略——记忆被注入 context 后，AI 在回复中不应透露"根据我的记忆"或"我记得你说过"等表述。配合边界示例（如"不要在无关上下文中提及用户已故宠物"），确保记忆应用既有效又不突兀。
- 结论: 记忆系统的用户体验质量取决于"隐性应用"而非"显性提及"。
案例 2: Claude Code 的文件型持久记忆
- 问题: 编程 agent 如何在长项目中保持上下文连贯性？
- 设计模式的使用: 采用文件型持久记忆，将记忆存储为 typed 文件（user/feedback/project/reference），并使用 MEMORY.md 作为索引文件。每次会话开始时通过读取 MEMORY.md 恢复项目上下文。这种方式比数据库存储更透明、更可编辑、更易于版本控制。
- 结论: 在专业工具场景中，记忆的透明性和可编辑性比自动化程度更重要。
案例 3: FlintK12 的教学记忆强制创建
- 问题: AI 如何在有限的对话轮次中积累足够的学生画像以实现个性化教学？
- 设计模式的使用: 实现 mandatory create_memory call——在每次教学交互中，AI 必须调用记忆创建工具记录学生的兴趣、理解水平和学习偏好。Pedagogical memory 存储兴趣、偏好、年级等信息，用于后续教学内容的个性化适配。
- 结论: 在高价值场景中（教育效果直接取决于个性化程度），强制创建记忆优于等待显式触发。
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.
- 8d ago First seen · 117 lines · 166 tokens per session scan A 8e7c88f4aefd
memory-system is a skill published in the GitHub repository kangarooking/system-prompt-skills (182 stars, last pushed 4mo ago), licensed MIT. It adds 166 tokens to every session and 2,515 once invoked, about $0.0008 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
memory-patterns
Use this skill when managing agent context and memory. Covers short-term vs long-term memory, context compaction, and persistent note-taking patterns.
shodh-memory
Persistent memory system for AI agents. Use this skill to remember context across conversations, recall relevant information, and build long-term knowledge. Activate when you need to store decisions, learnings, errors, or context that should persist beyond the current session.
kn-extract
Use when extracting reusable patterns, decisions, failures, or knowledge into documentation.
kn-init
Use at the start of a new session to read project docs, understand context, and see current state.
mnemo-cortex
Installs and wires Mnemo Cortex (local-first persistent memory) into OpenClaw and other MCP-capable agents. Use for cross-session recall, decision history, or multi-agent shared memory.
distill-feedback
A process for turning corrections you give an agent into lasting working rules. It reviews saved conversations and asks for approval before changing those rules.