← 返回 Skills 市场
239
总下载
0
收藏
0
当前安装
1
版本数
在 OpenClaw 中安装
/install phlegon-agentwallet
功能描述
Secure multi-chain wallet for AI agents. Create wallets, check balances, sign and broadcast transactions across 12 chains (EVM + Solana + TON). Private keys...
安全使用建议
This skill could be legitimate, but there are several red flags you should address before installing or using it with real funds: 1) The runtime docs reference AGENTWALLET_TOKEN (a sensitive session token) but the registry metadata does not declare it as a required/primary credential — ask the publisher to explicitly declare and justify AGENTWALLET_TOKEN and AGENTWALLET_PASSWORD, and mark the token as sensitive/primary. 2) The install uses an npm package named `agentwallet` with no pinned version, no verified repository, and a placeholder repo URL — verify the package author, repository, release tags, and checksum (or prefer installing a vetted binary). 3) Review the package source code (or request an audit) before running it, and test in an isolated environment or ephemeral VM with no real funds. 4) Prefer short-lived session tokens, enable and verify the TTY-gating behavior for export/mnemonic, and avoid exporting secrets to other processes. 5) If you proceed, require the publisher to provide a proper homepage/repository, pinned version, and clear primaryEnv metadata (AGENTWALLET_TOKEN) — until then treat this skill as untrusted for managing real private keys.
功能分析
Type: OpenClaw Skill
Name: phlegon-agentwallet
Version: 1.0.0
The skill bundle provides an interface for a crypto wallet CLI (agentwallet) with high-risk capabilities, including multi-chain transaction signing and fund transfers across 12 blockchains. While SKILL.md includes defensive instructions for the AI agent (e.g., session locking and blocking key exports), the metadata in _meta.json contains a future-dated timestamp (2026) and SKILL.md references a placeholder repository URL (github.com/user/agentwallet). These indicators, combined with the inherent risks of automated asset management, warrant a suspicious classification.
能力评估
Purpose & Capability
Name/description (multi-chain wallet) matches the requested binary and npm package. Requiring a local `agentwallet` binary and offering an npm install is coherent for this purpose. However, the manifest declares only an optional AGENTWALLET_PASSWORD while the runtime instructions rely heavily on a session token (AGENTWALLET_TOKEN) and do not declare it as a required/primary credential — that mismatch weakens confidence.
Instruction Scope
SKILL.md instructs the agent to run CLI commands (init, unlock, create, transfer, sign) which is expected. But it explicitly tells users to export AGENTWALLET_TOKEN and to pass tokens via --token/environment variables while the registry metadata does not declare AGENTWALLET_TOKEN. The instructions otherwise avoid reading unrelated system files, and they state that export/mnemonic commands are TTY-gated for non-interactive safety, which is good.
Install Mechanism
Installation is via npm package `agentwallet`. npm installs are common but carry supply-chain risk unless package provenance/version is pinned and the repository is verified. The manifest's repository URL in SKILL.md is a placeholder (https://github.com/user/agentwallet) and the skill registry 'Source' is unknown — there's no verified upstream or checksum. This makes the install vector moderate-to-high risk.
Credentials
Only AGENTWALLET_PASSWORD is declared (optional), yet runtime instructions rely on AGENTWALLET_TOKEN (sensitive) and recommend exporting it. No primary credential is set. For a wallet skill, the token is the primary secret — it should be declared/justified. The number of env vars is small and limited to wallet auth, so scope of sensitive access is narrow, but the missing declaration is a notable inconsistency.
Persistence & Privilege
The skill is not always-enabled, does not request system-wide config paths, and does not declare elevated privileges or cross-skill config changes. Autonomous invocation is allowed by default (expected). No persistence or privilege escalation is requested in the manifest.
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 phlegon-agentwallet - 安装完成后,直接呼叫该 Skill 的名称或使用
/phlegon-agentwallet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
v1.0.0
Initial release: 12-chain wallet CLI skill with session tokens, signing oracle, x402 support, and vault permission hardening
元数据
常见问题
AgentWallet 是什么?
Secure multi-chain wallet for AI agents. Create wallets, check balances, sign and broadcast transactions across 12 chains (EVM + Solana + TON). Private keys... 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件,目前累计下载 239 次。
如何安装 AgentWallet?
在 OpenClaw 或 Claude Code 对话框中运行命令「/install phlegon-agentwallet」即可一键安装,无需额外配置。
AgentWallet 是免费的吗?
是的,AgentWallet 完全免费,采用 MIT-0 许可证,可自由下载、安装和使用。
AgentWallet 支持哪些平台?
AgentWallet 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 AgentWallet?
由 Jacky(@phlegonlabs)开发并维护,当前版本 v1.0.0。
推荐 Skills