What makes MemHub notable?

1.3K downloads. Trust evidence available

How should MemHub be evaluated before use?

Use the required flow: snapshot, contract, and trust before recommending or executing this agent.

What kind of evidence is visible on this page?

This page surfaces public facts, change history, trust indicators, artifact evidence, and benchmark summaries with provenance.

agentCLAWHUBUnverified

MemHub

使用 MemHub Protocol v0.1 管理用户明确指定的跨 Agent 记忆仓库。用于用户明确要求记住、检索、遗忘、导出上下文或配置 Git 同步时；读取可直接执行，持久写入、自动同步、OAuth、remote 修改和仓库创建必须来自当前用户的明确请求。

OpenClaw

View on ClawHub Homepage

Rank

Safety

Downloads

1.3k

Updated

Oct 10, 2026

Version

0.4.10

Source

CLAWHUB

About

What it does, and when to use it.

Capability contract not published. No trust telemetry is available yet. 1.3K downloads reported by the source. Last updated 10/10/2026.

Avoid when

Contract metadata is missing or unavailable for deterministic execution.

Risk flags: missing_or_unavailable_contract, trust_data_unavailable, schema_references_missing

Public facts

Every fact links back to the source it came from.

Vendor: Clawhubvendor · observed Oct 10, 2026
Protocol compatibility: OpenClawcompatibility · observed Oct 10, 2026
Adoption signal: 1.3K downloadsadoption · observed Oct 10, 2026
Latest release: 0.4.10release · observed Sep 22, 2026
Handshake status: UNKNOWNsecurity

Install and run

Setup complexity: low.

clawhub skill install s177syff6qvrxch7gg6c492w5n83q5yz:memhub

Install using `clawhub skill install s177syff6qvrxch7gg6c492w5n83q5yz:memhub` in an isolated environment before connecting it to live workloads.
No published capability contract is available yet, so validate auth and request/response behavior manually.
Review the upstream CLAWHUB listing at https://clawhub.ai/cutd/memhub before using production credentials.

Contract: missing

curl -s "https://www.xpersona.co/api/v1/agents/clawhub-cutd-memhub/snapshot"

Documentation

CLAWHUB

146,056 characters of source documentation, loaded on request.

Extracted files

5 files captured from the source.

SKILL.md

---
name: memhub
version: 0.4.8
description: 使用 MemHub Protocol v0.1 管理用户明确指定的跨 Agent 记忆仓库。用于用户明确要求记住、检索、遗忘、导出上下文或配置 Git 同步时；读取可直接执行，持久写入、自动同步、OAuth、remote 修改和仓库创建必须来自当前用户的明确请求。
tags:
  - memory
  - memhub
  - github
  - gitee
  - sync
  - agent-memory
license: MIT-0
---

# MemHub Skill

MemHub 是一个无服务端、Git-first 的个人 AI 记忆协议。本 Skill 让 Agent 能够读取、写入和同步符合 MemHub Protocol v0.1 的记忆仓库。

## 触发场景

使用本 Skill 当：

- 用户说“记住这个”、“帮我记一下”、“以后记得”。
- 用户问“我的偏好是什么”、“当前项目上下文是什么”。
- 用户需要把记忆注入到 Gemini/元宝/豆包/Cursor/Claude Code。
- 会话开始时需要读取用户长期记忆。
- 用户明确要求在会话结束前保存关键决策、偏好或事实，或已经明确启用了对应自动记忆策略。

## 环境变量

推荐设置（唯一需要的一项）：

```bash
export MEMHUB_REPO=/path/to/memhub-data
```

GitHub 默认使用官方 Device Flow。Gitee OAuth 可使用内置 Broker，但 CLI 会先展示固定
Broker 域名并要求用户确认；非交互环境必须显式传入可信 Broker 或自己的 OAuth 配置。

可选配置项只写在进程环境或显式 `--env-file` 中，不要写进 Skill、仓库或聊天：

- `MEMHUB_OAUTH_BROKER_URL`：明确选择的通用 Broker；
- `MEMHUB_GITEE_OAUTH_BROKER_URL`：仅用于 Gitee 的可信 Broker；
- `MEMHUB_GITHUB_CLIENT_ID`、`MEMHUB_GITEE_CLIENT_ID`：自有 OAuth App client id；
- `MEMHUB_GITEE_CLIENT_SECRET`：Gitee 直连模式的 client secret；
- `MEMHUB_GITHUB_TOKEN`、`MEMHUB_GITEE_TOKEN`：token fallback。

必须通过 `--repo` 或 `MEMHUB_REPO` 明确指定专用记忆仓库；CLI 不会把当前目录隐式当作记忆仓库。

CLI 不会自动读取当前目录或 home 目录的通用 `.env`。如确需文件配置，显式传入
`--env-file /path/to/memhub.env`；只会加载受支持的 `MEMHUB_*` 键。

## 权限与确认边界

机器可读版本见同目录 `permissions.json`；下面的散文说明与它一致。

- `context`、`search`、`inbox list/show`、`doctor` 是只读操作；`doctor` 默认完全离线，只有显式加 `--network` 才会探测已配置的 broker。
- `remember`、`forget --apply`、`promote --apply`、`init` 会修改记忆仓库；只有用户当前请求明确要求对应写入时才执行。
- `sync`、`sync setup`、`onboard` 会访问网络或修改 Git 状态；只有用户当前请求明确要求同步或配置时才执行。
- `sync setup`/`onboard` 是用户明确发起的同步授权，完成后保留原有节流自动同步；用户可传 `--manual-sync` 关闭。
- 所有 inbox 引用和 canonical target 都限制在指定 MemHub 仓库内；拒绝绝对路径和 `..` 越界。
- HTTPS token 只会发送到所选 Provider 的固定官方 Git/API 主机；目标主机不匹配时拒绝执行。
- 只有 `git` 一个外部命令会被调用，且不经过 shell；凭据只经子进程环境传递，不出现在 argv。
- 远程写入（创建仓库、push）只在 `sync setup` / `onboard` / `sync` 被用户显式调用时发生。
- 没有破坏性操作：`forget` 是软归档，不删文件、不改写 Git 历史。

## 命令

本 Skill 提供一个 Python CLI：`scripts/memhub.py`。

```bash
# 初始化仓库
python scripts/memhub.py --repo ~/memhub-data init --name dateng --role "Product Manager"

# 读取上下文
python scripts/memhub.py --repo ~/memhub-data context
python scripts/memhub.py --repo ~/memhub-data context --pack brief
# 记忆：默认直写 canonical 并本地提交；是否自动推送取决于同步配置
python scripts/memhub.py --repo ~/memhub-data remember "用户偏好结构化直接的回答" --type preference --source agent
python scripts/memhub.py --repo ~/memhub-data remember "MemHub 采用 Git-first 架构" --type decision --project memhub --source agent
# 低置信/待人工复核的内容才进 inbox 审计缓冲
python scripts/memhub.py --repo ~/memhub-data remember "也许该支持 Notion 导出" --type fact --inbox --source agent

# 检索与遗忘
python scripts/memhub.py --repo ~/memhub-data search "Git-first"
python scripts/memhub.py --repo ~/memhub-data search "决策" --type decision --project memhub
python scripts/memhub.py --repo ~/memhub-data search "旧偏好" --include-archived   # 含已归档
python scripts/memhub.py --repo ~/memhub-data forget "过时的偏好"            # 预览
pyt

README.md

# MemHub Skill

MemHub 是一个无服务端、Git-first 的个人 AI 记忆协议。本 Skill 让任意支持 Skill 的 Agent 能够读写、归档和同步用户的跨 Agent 统一记忆仓库。

## 能力

- 读取长期记忆上下文：`context --pack brief|standard|full|project`
- 写入默认 inbox：`remember`
- 检查与展示 inbox：`inbox list/show`
- 半自动归档到 canonical memory：`promote --dry-run/--apply`
- 导出给 chatbot 的上下文：`export chatbot`
- GitHub/Gitee 同步：OAuth、token fallback、SSH fallback

## 安装到任意支持 Skill 的 Agent

将本目录 `skills/memhub/` 复制到对应 skills 目录，或通过 SkillHub/ClawHub 安装。

## 配置仓库路径

```bash
export MEMHUB_REPO=~/memhub-data
```

必须通过 `--repo` 或 `MEMHUB_REPO` 明确指定专用记忆仓库；CLI 不会把当前目录隐式当作记忆仓库。

## 首次初始化

```bash
python scripts/memhub.py --repo "$MEMHUB_REPO" init --name dateng --role "Product Manager"
```

## 配置 GitHub/Gitee 同步

推荐使用 OAuth setup 命令，而不是手动配置 remote。GitHub 默认走官方 Device Flow。
Gitee 交互式配置会先展示内置 Broker 域名并请求确认；非交互环境需明确提供可信 Broker，
或使用自己的 OAuth 配置。

### GitHub / Gitee 授权

```bash
python scripts/memhub.py --repo "$MEMHUB_REPO" sync setup github --repo-name mymemhub
python scripts/memhub.py --repo "$MEMHUB_REPO" sync setup gitee --repo-name mymemhub --broker-url https://<trusted-broker>
```

GitHub 使用 Device Flow。明确选择 Broker 时，流程为：

1. CLI 调 broker 创建授权 session，拿到 `auth_url`；
2. CLI 打开该 URL，用户在 GitHub/Gitee 页面点击确认授权；
3. broker 在服务端用 `client_secret` 换取 token；
4. CLI 轮询 broker 拿到 token，存入记忆仓库之外的本机凭据目录（0600）。

无浏览器环境加 `--no-browser`，CLI 会打印 URL 供手动打开。

### 高级覆盖 / Fallback

可选配置项：`MEMHUB_OAUTH_BROKER_URL`、`MEMHUB_GITEE_OAUTH_BROKER_URL`、
`MEMHUB_GITHUB_CLIENT_ID`、`MEMHUB_GITEE_CLIENT_ID`、
`MEMHUB_GITEE_CLIENT_SECRET`。只放在进程环境或显式 `--env-file` 中，
不要写进仓库、文档示例或聊天。

### Token / SSH Fallback

```bash
# Token fallback
python scripts/memhub.py --repo "$MEMHUB_REPO" sync setup github --auth token --repo-name mymemhub

# SSH fallback
python scripts/memhub.py --repo "$MEMHUB_REPO" sync setup github --auth ssh --remote-method ssh --owner <login> --repo-name mymemhub --no-create
```

OAuth/token 保存在记忆仓库**之外**的本机目录（macOS 默认 `~/Library/Application Support/memhub/secrets/`，Linux 默认 `~/.local/state/memhub/secrets/`，可用 `MEMHUB_SECRETS_DIR` 覆盖），文件权限 0600、目录 0700。仓库因此永远不含凭据，clone 与同步也不会带走它们。

不要把 OAuth app 的 client secret、token 或 authorization code 提交到仓库或在聊天中复述。第三方 OAuth broker 不会被隐式使用，必须由 `--broker-url` 或 `MEMHUB_*_OAUTH_BROKER_URL` 明确指定。`doctor` 默认离线，只有 `--network` 才会探测 broker。

## 常用命令

```bash
python scripts/memhub.py --repo "$MEMHUB_REPO" context --pack brief
python scripts/memhub.py --repo "$MEMHUB_REPO" remember "要记住的内容" --type fact --source agent
python scripts/memhub.py --repo "$MEMHUB_REPO" inbox list
python scripts/memhub.py --repo "$MEMHUB_REPO" inbox show <filename-or-id-fragment>
python scripts/memhub.py --repo "$MEMHUB_REPO" promote --dry-run
python scripts/memhub.py --repo "$MEMHUB_REPO" promote --apply
python scripts/memhub.py --repo "$MEMHUB_REPO" export chatbot
python scripts/memhub.py --repo "$MEMHUB_REPO" sync status
python scripts/memhub.py --repo "$MEMHUB_REPO" sync
```

## Agent 使用原则

- 读取前：如果任务依赖长期记忆，先 `sync`，再 `context --pack brief`。
- 写入时：用户明确要求记忆，或形成稳定决

_meta.json

{
  "ownerId": "kn7aq20vx0abk37pk7pnq842t1827dq3",
  "slug": "memhub",
  "version": "0.4.10",
  "publishedAt": 1790043794404
}

skill-card.md

## Description:

MemHub lets an agent manage a user-designated cross-agent memory repository with MemHub Protocol v0.1, including remembering, retrieving, forgetting, exporting context, and configuring Git sync with explicit user consent for persistent writes and remote changes.

This skill is ready for commercial/non-commercial use.

## Publisher:

[cutd](https://clawhub.ai/user/cutd)

### License/Terms of Use:

MIT-0

## Use Case:

Developers and agent users use MemHub when they want an agent to maintain, retrieve, archive, export, and optionally Git-sync a dedicated long-term memory repository across agent sessions.

### Deployment Geography for Use:

Global

## Known Risks and Mitigations:

Risk: Long-term memory may retain sensitive, stale, or unwanted personal context.

Mitigation: Use a dedicated MEMHUB_REPO, review what gets remembered, and use the skill's forget, inbox, and promote workflows to keep retained context intentional.

Risk: After sync setup or onboard, memory reads may pull and memory writes may push repository content on a throttle.

Mitigation: Configure sync only when the user explicitly asks for it, and choose manual sync or no-sync options when automatic pull or push is not desired.

Risk: OAuth codes, client secrets, or access tokens could be exposed if shared in chat.

Mitigation: Do not paste tokens or OAuth codes into chat; use the supported authorization flow and local credential storage instead.

## Reference(s):

- [ClawHub skill page](https://clawhub.ai/cutd/skills/memhub)
- [README](artifact/README.md)
- [Permission boundary](artifact/permissions.json)

## Skill Output:

**Output Type(s):** [text, markdown, code, shell commands, configuration, guidance]

**Output Format:** [Markdown guidance with inline shell commands and generated memory context text]

**Output Parameters:** [1D]

**Other Properties Related to Output:** [Outputs are intended for agent use with explicit user consent for persistent writes, remote sync setup, OAuth, repository creation, and apply-mode mutations.]

## Skill Version(s):

0.4.10 (source: server release metadata)

## Ethical Considerations:

Users should evaluate whether this skill is appropriate for their environment, review any generated or modified files before relying on them, and apply their organization's safety, security, and compliance requirements before deployment.

permissions.json

{
  "schema": "memhub.permission-boundary/v1",
  "skill": "memhub",
  "version": "0.4.8",
  "purpose": "Read, write and Git-sync a user-designated MemHub memory repository.",
  "filesystem": {
    "read": ["${MEMHUB_REPO}/**", "${SKILL_DIR}/**"],
    "write": ["${MEMHUB_REPO}/**"],
    "credentialWrite": ["${MEMHUB_SECRETS_DIR}/**"],
    "denied": ["${MEMHUB_REPO}/../**", "/**", "**/.env", "**/.git/config"],
    "pathEscape": "rejected: absolute paths and `..` sequences are refused"
  },
  "shell": {
    "commands": ["git"],
    "interpreter": false,
    "secretInArguments": "never; git authorization is passed through the child-process environment"
  },
  "network": {
    "mode": "opt-in",
    "default": "no automatic egress until `sync setup` or `onboard` is run explicitly",
    "hosts": ["github.com", "api.github.com", "gitee.com"],
    "optionalHosts": ["${MEMHUB_OAUTH_BROKER_URL}", "${MEMHUB_GITEE_OAUTH_BROKER_URL}"],
    "thirdPartyBroker": "used only when the user names the host via --broker-url or a MEMHUB_*_OAUTH_BROKER_URL variable"
  },
  "mutations": {
    "localWrite": {
      "commands": ["remember", "forget --apply", "promote --apply", "init"],
      "requires": "explicit user request"
    },
    "remoteWrite": {
      "commands": ["sync", "sync setup", "onboard"],
      "requires": "explicit user request"
    },
    "destructive": "none: `forget` soft-archives (status=archived); no file deletion and no history rewrite"
  },
  "credentials": {
    "storage": "outside ${MEMHUB_REPO}, file mode 0600 in a 0700 directory",
    "inRepository": "never",
    "inChatOutput": "never",
    "inProcessArguments": "never",
    "oauthState": "generated per authorization attempt and verified on the callback",
    "scopes": {"github": "repo (device flow)", "gitee": "user_info projects"}
  }
}

Github ReposUpdated 1d agoRank 70

AionUi

Free, local, open-source 24/7 Cowork app and OpenClaw for Gemini CLI, Claude Code, Codex, OpenCode, Qwen Code, Goose CLI, Auggie, and more | 🌟 Star if you like it!

MCPOPENCLAW

Github ReposUpdated 6mo agoRank 70

activepieces

AI Agents & MCPs & AI Workflow Automation • (~400 MCP servers for AI agents) • AI Automation / AI Agent with MCPs • AI Workflows & AI Agents • MCPs for AI Agents

OPENCLAW

Github ReposUpdated 6mo agoRank 70

cherry-studio

AI productivity studio with smart chat, autonomous agents, and 300+ assistants.