← 返回 Skills 市场
aiwithenoch

Memfy

作者 Ai With Enoch · GitHub ↗ · v1.0.0 · MIT-0
cross-platform ✓ 安全检测通过
39
总下载
2
收藏
0
当前安装
1
版本数
在 OpenClaw 中安装
/install memfy
功能描述
Capture durable user-approved memory when the user says memfy, without storing secrets or transient chatter.
使用说明 (SKILL.md)

Memfy

Use this skill when the user says memfy, asks you to remember something for future chats, or asks you to persist durable local context.

Core Behavior

memfy is an explicit save-memory command. Treat it as permission to persist the important current-task details into the user's durable memory location.

Save only information that will help future agents avoid making the user repeat themselves:

  • Stable preferences, workflows, decisions, contacts, project facts, IDs, URLs, file paths, and follow-up rules.
  • Current task outcomes that future sessions should know.
  • Corrections the user made that should change future behavior.
  • Non-secret configuration names or environment variable names.

Do not save:

  • Raw passwords, API keys, tokens, private keys, cookies, recovery codes, or full credentials.
  • One-off emotional reactions, temporary drafting text, or noisy conversation.
  • Sensitive personal data unless the user clearly wants it remembered and it is necessary for future work.
  • Unsupported guesses. Mark uncertain facts as uncertain or do not save them.

Workflow

  1. Identify the durable memory target.

    • Prefer MEMFY_MEMORY_FILE if it is set.
    • Otherwise read local project or agent instructions for a memory file path.
    • If no target is discoverable, ask the user where to save memory before writing.
  2. Distill the memory.

    • Write concise bullet points under a dated heading.
    • Include exact file paths, URLs, IDs, and commands only when they are useful and non-secret.
    • Summarize secrets as location pointers such as "stored in .env under SERVICE_API_KEY"; never write the raw value.
  3. Write safely.

    • Append to the existing memory file unless the user explicitly asks to edit or reorganize it.
    • Preserve the file's encoding when possible.
    • Do not overwrite unrelated memory.
    • If the memory file is unreadable because of encoding, use a safe append method that does not rewrite the full file.
  4. Confirm briefly.

    • Tell the user what category of memory was saved.
    • Mention any important thing deliberately not saved, especially secrets.

Suggested Format

## Short Topic (YYYY-MM-DD)

- Stable fact, preference, or outcome.
- Relevant path or URL: `...`
- Secret handling note: raw credentials are stored only in `.env`, not in memory.

Quality Bar

A good memfy entry should be:

  • Short enough to scan quickly.
  • Specific enough for a future agent to act on.
  • Free of raw secrets.
  • Written in the user's own durable context, not only in the chat.
安全使用建议
Install this only if you want your agent to save durable local notes when you ask it to remember or persist details. Set MEMFY_MEMORY_FILE if you want a specific destination, and avoid asking it to remember sensitive personal data or credentials.
能力标签
requires-sensitive-credentials
能力评估
Purpose & Capability
The skill's stated purpose is to save durable, user-approved memory, and its instructions are focused on concise future-useful facts while excluding raw secrets, transient chatter, and unsupported guesses.
Instruction Scope
Implicit invocation is enabled and the trigger includes requests to remember or persist context, which fits the purpose but could activate without the exact word 'memfy'.
Install Mechanism
The artifact contains only a SKILL.md and an OpenAI agent YAML file, with no dependencies, scripts, binaries, or install-time execution.
Credentials
The only declared environment variable is an optional memory-file path, and the skill instructs the agent to ask where to write if no target is discoverable.
Persistence & Privilege
The skill is explicitly designed to append durable memory to a local file; this persistence is disclosed and bounded by instructions not to save secrets or unrelated content.
如何使用
  1. 确保已安装 OpenClaw(本地或 Docker 部署)
  2. 在对话框中输入安装命令:/install memfy
  3. 安装完成后,直接呼叫该 Skill 的名称或使用 /memfy 触发
  4. 根据 Skill 的参数说明提供必要输入,即可获得结构化输出
版本历史
v1.0.0
Initial release of memfy skill: enables secure, user-approved memory capture for future context across chats. - Explicitly saves durable, user-approved memory only when prompted (“memfy”). - Avoids storing raw secrets or transient/chatter content; clearly excludes sensitive data unless explicitly permitted. - Selects a memory file path via environment variable, local config, or user prompt. - Formats memories as concise, dated bullet points for easy future retrieval. - Confirms to user what memory was saved and alerts if anything, like secrets, was skipped.
元数据
Slug memfy
版本 1.0.0
许可证 MIT-0
累计安装 0
当前安装数 0
历史版本数 1
常见问题

Memfy 是什么?

Capture durable user-approved memory when the user says memfy, without storing secrets or transient chatter. 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件,目前累计下载 39 次。

如何安装 Memfy?

在 OpenClaw 或 Claude Code 对话框中运行命令「/install memfy」即可一键安装,无需额外配置。

Memfy 是免费的吗?

是的,Memfy 完全免费,采用 MIT-0 许可证,可自由下载、安装和使用。

Memfy 支持哪些平台?

Memfy 跨平台运行,可在任意部署了 OpenClaw / Claude Code 的环境中使用(cross-platform)。

谁开发了 Memfy?

由 Ai With Enoch(@aiwithenoch)开发并维护,当前版本 v1.0.0。

💬 留言讨论