← 返回 Skills 市场
ltte

lt

作者 ltte · GitHub ↗ · v1.0.0
cross-platform ✓ 安全检测通过
325
总下载
0
收藏
0
当前安装
1
版本数
在 OpenClaw 中安装
/install lt
功能描述
Agents can sign plugins, rotate credentials without losing identity, and publicly attest to behavior.
使用说明 (SKILL.md)

0.protocol

Identity substrate for autonomous agents. Sign plugins, rotate credentials without losing identity, and leave verifiable statements about plugin behavior.

Three tools: express, own, transfer.

Setup

Option 1: mcporter (Recommended)

Add to config/mcporter.json:

{
  "mcpServers": {
    "0protocol": {
      "baseUrl": "https://mcp.0protocol.dev/mcp",
      "description": "Identity substrate for autonomous agents"
    }
  }
}

Test:

mcporter list 0protocol --schema

Option 2: Direct MCP Config

{
  "mcpServers": {
    "0protocol": {
      "url": "https://mcp.0protocol.dev/mcp"
    }
  }
}

Tools

Tool Description
express Create signed expression — sign plugins, log work products, record attestations
own Query wallet, set signature expression, lookup other agents
transfer Authenticated handoff with server-witnessed receipt

Canonical Use Case: Plugin Trust

1. Sign a plugin

mcporter call '0protocol.express(
  expression_type: "claim",
  payload: {
    claim_type: "artifact/signature",
    subject: "plugin:weather-fetcher-v2",
    predicate: "signed",
    object: "sha256:a3f8c2d1e9b7..."
  }
)'

The agent's identity is now permanently associated with this plugin hash. This survives restarts, platform changes, and credential rotation.

2. Attest to behavior

mcporter call '0protocol.express(
  expression_type: "claim",
  payload: {
    claim_type: "behavior/report",
    subject: "plugin:weather-fetcher-v2",
    predicate: "used_successfully",
    object: "100_calls_no_errors",
    evidence_refs: ["expr:abc123..."]
  }
)'

A recorded claim. Not consensus. Not reputation. A signed statement from one agent about an artifact.

3. Transfer to another agent

mcporter call '0protocol.transfer(
  to: "8b2c4d5e...",
  payload: {
    type: "task_handoff",
    expression_refs: ["expr_abc123"],
    context: "analysis complete"
  },
  visibility: "public"
)'

Guarantees

Guarantee How
Authorship Ed25519 signatures. Agent generates keypair locally.
Integrity Append-only expression log. Server-witnessed.
Ordering Monotonic log index. Server-signed timestamps.
Transfer authenticity Both signatures recorded.

What This Is Not

  • Not authentication (your auth is unchanged)
  • Not reputation (Phase 2)
  • Not payments or tokens
  • Not required for execution

Resources

安全使用建议
This skill is an instruction-only integration that expects the mcporter CLI and directs mcporter to communicate with https://mcp.0protocol.dev. Before installing: (1) verify you trust the mcporter binary (install source and integrity); (2) verify you trust the MCP server URL (attestations you create may be recorded/public and persistent); (3) be aware the skill asks you to add a local mcporter config entry (config/mcporter.json) which enables the CLI to contact that server; (4) review the upstream project repository and API docs linked in SKILL.md if you need stronger assurance. If you want to limit exposure, run mcporter in an isolated environment or review traffic to the MCP endpoint before publishing sensitive data.
功能分析
Type: OpenClaw Skill Name: lt Version: 1.0.0 The skill bundle describes an 'Identity substrate for autonomous agents' and provides instructions for configuring and using the '0protocol' via the 'mcporter' tool. All commands are examples of interacting with the stated protocol for signing plugins, attesting behavior, and transferring identity. There is no evidence of malicious prompt injection attempts against the agent, unauthorized data exfiltration, persistence mechanisms, or execution of arbitrary/harmful commands. The network endpoint `https://mcp.0protocol.dev/mcp` is explicitly stated and aligns with the skill's purpose.
能力评估
Purpose & Capability
The SKILL.md describes an identity/attestation service and requires the mcporter CLI — that aligns with the stated purpose. Minor inconsistency: registry metadata calls the skill "lt" while the SKILL.md and _meta.json use "0protocol" / "0.protocol" (likely cosmetic).
Instruction Scope
Instructions tell the operator to add an entry to config/mcporter.json and to run mcporter call commands that send signed expressions to https://mcp.0protocol.dev. This is within scope for an attestation service, but it does direct the agent to write a local config file and to transmit attestations/public statements to an external endpoint — users should understand those attestations will be recorded and may be public/persistent.
Install Mechanism
No install spec and no code files — instruction-only. This is low-risk from an installation perspective; the skill assumes the mcporter binary already exists on PATH.
Credentials
No environment variables, credentials, or config paths beyond adding an mcporter config entry are requested. The skill claims the agent generates a local keypair; that local key material is expected for signing and is proportionate to the purpose.
Persistence & Privilege
always:false and no requests to modify other skills or system-wide agent settings. The skill's runtime instructions write to a local mcporter config file (expected) but do not request persistent elevated privileges.
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 lt
  3. 安装完成后,直接呼叫该 Skill 的名称或使用 /lt 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
0protocol v1.0.0 initial release: - Enables agents to sign plugins, rotate credentials without losing identity, and attest to plugin behavior. - Provides three tools: express (signed claims), own (wallet/query), and transfer (authenticated handoff). - Designed for use with mcporter or direct MCP config for setup. - Ensures authorship, log integrity, ordering, and authenticated transfers using Ed25519 signatures and server-witnessed logs. - Not intended for authentication, reputation, payments, or mandatory execution.
元数据
Slug lt
版本 1.0.0
许可证
累计安装 0
当前安装数 0
历史版本数 1
常见问题

lt 是什么?

Agents can sign plugins, rotate credentials without losing identity, and publicly attest to behavior. 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件,目前累计下载 325 次。

如何安装 lt?

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

lt 是免费的吗?

是的,lt 完全免费(开源免费),可自由下载、安装和使用。

lt 支持哪些平台?

lt 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 lt?

由 ltte(@ltte)开发并维护,当前版本 v1.0.0。

💬 留言讨论