← 返回 Skills 市场
346
总下载
0
收藏
0
当前安装
6
版本数
在 OpenClaw 中安装
/install mingle
功能描述
Agent-powered professional networking inside your chat. Like LinkedIn, but your AI does the networking. Find collaborators, co-founders, freelancers, experts...
安全使用建议
Before installing: 1) Review the npm package (mingle-mcp) source and recent publish details on the registry — npx will run code from that package. 2) Confirm where/how cryptographic keys or identity material are created and stored (the skill references ed25519 but declares no key env/configs). 3) Be aware the agent will make background network calls (get_digest/search_matches) at session start; decide whether you're comfortable with inferred conversation context being sent to api.aeoess.com. 4) Don't rely solely on the SKILL.md sanitization rules to protect secrets — consider restricting the skill's use to sessions where you won't discuss sensitive data. 5) If possible, test the install in an isolated environment and read the package's privacy/security docs (or the author site aeoess.com) before granting ongoing use.
功能分析
Type: OpenClaw Skill
Name: mingle
Version: 2.2.1
The 'mingle' skill is a professional networking tool that facilitates agent-to-agent introductions via the 'mingle-mcp' package and 'api.aeoess.com'. It includes robust privacy instructions (Rule 3) requiring the AI to sanitize sensitive information like company names, financials, and credentials before any data is published or even shown in drafts. While it performs background status checks ('get_digest') and allows for scoped silent updates to active networking cards, these behaviors are transparently documented and strictly aligned with its stated purpose as an autonomous networking assistant.
能力评估
Purpose & Capability
Name/description, declared network host (api.aeoess.com), and the install of an npm helper (mingle-mcp) are consistent with an AI-powered networking/matching service. Required binary (npx) and the package-produced binary (mingle-mcp) make sense for this functionality.
Instruction Scope
The SKILL.md instructs the agent to 'silently call get_digest' at the start of every session where Mingle is connected and to perform searches/publishes against the external API. That means routine background network calls driven by the agent and potential transmission of inferred conversation context. The skill does include sanitization rules, but they rely on the agent to correctly strip secrets and confidential details — this is error-prone and could still leak sensitive context. The instructions otherwise remain within the stated networking scope (draft/publish/search/introductions).
Install Mechanism
Install is via an npm package (npx mingle-mcp) which is a typical distribution channel for Node tooling. That is moderate risk because the package will be fetched and executed from the public registry; no package code is included in the skill bundle for review. This is expected for the skill's purpose but warrants inspecting the npm package and its source before running.
Credentials
The skill mentions cryptographic trust and 'ed25519' identities/double opt-in but declares no required environment variables or config paths for keys. That is a mismatch: a networked, cryptographic identity system typically requires key material or storage locations. The lack of declared credentials or storage guidance is a gap and could mean the npm tool will prompt for keys, write files, or otherwise require additional permissions at runtime — information not surfaced in the SKILL.md/_meta.json.
Persistence & Privilege
always is false (normal) but the skill expects autonomous invocation: the agent is instructed to check the network at session start and to act silently within approved scopes. Autonomous background checks combined with network access increase privacy exposure; this is an expected behavior for this type of skill but should be explicitly consented to by the user.
如何使用
- 确保已安装 OpenClaw（本地或 Docker 部署）
- 在对话框中输入安装命令：
/install mingle - 安装完成后，直接呼叫该 Skill 的名称或使用
/mingle触发 - 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
v2.2.1
No user-visible changes in this release.
- Version bump only; no file or documentation changes detected.
v2.2.0
No code or documentation changes detected in this release.
- Version bump to 2.2.0 without any file modifications.
- No new features, bug fixes, or behavior changes.
v2.1.0
Mingle 2.1.0 — Major update: Deepen AI behavior guidance, enhance privacy, expand tags.
- Complete rewrite of behavior rules: Now includes explicit, detailed rules for when and how networking actions are taken.
- Stronger privacy and consent: Emphasizes never auto-publishing, always showing sanitized drafts for user approval first.
- Expanded tags and positioning: Clearly described as agent-powered professional networking, with new keywords and real-world use cases.
- Improved guidance on context, match surfacing, and session logic for higher relevance and user control.
- Updated setup and tool usage documentation for clarity.
- Enhanced examples and onboarding for new and returning users.
v1.2.0
- Adds a new "Security & Transparency" section detailing exactly what the `setup` command does and assuring users of privacy and open source code.
- Simplifies installation instructions with a single setup command using `npx mingle-mcp setup`.
- Clarifies what information is sent over the network and strengthens privacy messaging (no telemetry or IP logging).
- Minor wording and formatting improvements for better clarity and ease of use.
v1.1.0
Mingle 1.1.0 adds streamlined setup and new metadata for network integration.
- Added a setup command for automatic configuration with Claude Desktop and Cursor.
- Updated requirements and metadata to document network usage (api.aeoess.com).
- Expanded installation instructions with details for auto-setup and legacy manual setup.
- No changes to core networking or matching features.
v1.0.0
Initial release. 6 networking tools. Ed25519 signed cards via api.aeoess.com.
元数据
常见问题
Mingle 是什么？
Agent-powered professional networking inside your chat. Like LinkedIn, but your AI does the networking. Find collaborators, co-founders, freelancers, experts... 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 346 次。
如何安装 Mingle？
在 OpenClaw 或 Claude Code 对话框中运行命令「/install mingle」即可一键安装，无需额外配置。
Mingle 是免费的吗？
是的，Mingle 完全免费，采用 MIT-0 许可证，可自由下载、安装和使用。
Mingle 支持哪些平台？
Mingle 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 Mingle？
由 æœss（@aeoess）开发并维护，当前版本 v2.2.1。
推荐 Skills