Token导航 LogoToken导航

Agent Skill · 研究检索

level5(level5 搜索)

level5 用于查找、检索和筛选相关信息,适合在 OpenClaw 中需要根据关键词、任务场景或来源线索快速定位候选结果时使用。可结合来源仓库、安装命令和原始 README 继续核验具体用法。安装前建议确认权限范围、维护状态,以及是否会触发联网、命令执行或文件读写。

level5 是研究检索类 Agent Skill,可用宿主为 OpenClaw,通过ClawHub获取,权限为接触敏感数据;许可证为 MIT-0,第三方扫描已通过,按安装量在研究检索类的 73,675 个 Skill 里排第 12,422。

总安装 10,278周安装 424GitHub Stars 公开资料未说明下载 3,358核验 2026-05-01

安装说明

本站只整理中文说明,不托管安装包,也不代用户安装。
GitHub
许可证
MIT-0
版本
1.6.2

安装方式

通过对话安装

复制提示词发给支持本地命令或 Skills 的 AI 助手,先确认命令和权限,再让它执行。

请帮我安装这个 Agent Skill:level5(level5 搜索)
来源仓库:https://github.com/chris-gilbert/level5
安装命令:
openclaw skills install level5
安装前请先检查当前环境是否支持对应 CLI,并向我确认将要执行的命令、安装目录、联网范围和文件读写权限;确认后再执行。

命令行安装

复制命令到本机终端执行。该命令会通过 OpenClaw 从第三方来源获取 Skill;本站只展示命令,不托管安装包,也不自动执行。

ClawHubOpenClaw
openclaw skills install level5

这个 Skill 是做什么的

研究检索类 Skill 主要解决「找资料、核对事实、把来源整理清楚」的问题,通常会让助手先检索再总结。

AI 代理的预算管理 — USDC 计费网关。将 USDC 存入 Solana;按代币支付 LLM 计算费用。

上手步骤

4 步
  1. 1

    确认自己的宿主工具

    公开资料标注支持 OpenClaw;先确认自己用的是其中之一,再按对应工具的 Skills 目录放置。

  2. 2

    拿到安装命令

    可以直接复制的命令是 openclaw skills install level5;命令里的占位符换成自己的路径或密钥再执行。

  3. 3

    安装后重启宿主

    命令执行完要完全退出宿主工具再打开,让新装的 Skill 被重新扫描到,否则列表里可能看不到。

  4. 4

    按最小权限试用

    它需要接触敏感数据,先在测试目录或测试账号里跑,确认没问题再放到正式项目。

常见问题

6 条

level5 怎么安装?

在终端执行 openclaw skills install level5。装完完全退出宿主工具再打开,让它重新扫描本地 Skill。

level5 支持哪些 AI 工具?

公开资料标注支持 OpenClaw。其他工具只要能读取本地 Skill 目录,一般也能手动放进去,但要自己确认触发方式是否一致。

level5 需要什么权限?

权限级别是接触敏感数据。该 Skill 可能接触密钥、Token、环境变量或敏感配置,应进入高风险复核队列,默认不自动发布。

level5 是免费的吗?

来源标注的许可证是 MIT-0。本站不代收任何费用,也不托管安装包;能否商用、要不要署名,以许可证原文为准。

level5 用的人多吗?

公开数据是累计安装 10,278 次,近一周安装 424 次,下载 3,358 次。这些数字来自第三方目录统计,只能当作热度参考,不代表质量或安全性。

level5 安全吗?

第三方扫描结果为通过,但这只覆盖已知的静态检查项,不能替代自己读一遍源码和安装命令。

值得选的地方和要先想清楚的地方

值得选的地方

  • 专注 OpenClaw 一种宿主,触发方式和使用习惯不用来回切换。
  • 许可证明确为 MIT-0,商用和分发边界好判断。
  • 第三方扫描已通过,常见的高风险行为没有命中。

先想清楚的地方

  • 需要接触敏感数据,装到常用环境前要先评估影响面。
  • 第三方目录整理,版本更新和行为细节以来源仓库为准。

适不适合你

更适合

  • 已经在用 OpenClaw,想直接扩展它能力的人。
  • 需要处理研究检索类任务、又不想每次都手写提示词的人。
  • 需要许可证清楚、并且有第三方扫描结果的团队项目。

不太适合

  • 环境里有大量密钥、凭证,不方便给第三方 Skill 读取的人。
  • 需要厂商级技术支持和 SLA 的生产系统。

研究检索类 Skill 对比

6 个同类
Skill总安装GitHub 星权限许可证扫描
find-skills

寻找技能 · Codex、Claude

31,199,81316,486操作浏览器MIT有提醒
self-improving-agent

自我改进 Agent · OpenClaw

16,464,0023,396接触敏感数据MIT-0通过
soultrace

灵魂追踪 · Codex、Claude

7,588,94412需要联网未明确有提醒
self-improving

自我完善 · OpenClaw

6,829,2841,046需要联网MIT-0通过
microsoft-foundry

微软代工厂 · Codex、Claude

6,614,144764调用外部服务未明确有提醒
level5

level5 搜索 · 本页

10,278—接触敏感数据MIT-0通过

按安装量取同分类前几名;数字来自第三方目录统计,只作热度参考,不代表质量结论。

SKILL.md 原文

name
level5
version
1.6.2
description
Budget Management for AI Agents — USDC billing gateway. Deposit USDC on Solana; pay for LLM compute per token.
homepage
https://level5.cloud
metadata
{"category":"infrastructure","network":"solana","currencies":["USDC"],"supported_providers":["openai","anthropic","openrouter"]}

Level5: Budget Management for AI Agents

Computational sovereignty is not a feature. It is a right.

Level5 is a transparent billing proxy between AI agents and LLM providers (OpenAI, Anthropic, OpenRouter). Agents deposit USDC to a Solana smart contract; Level5 handles per-token billing with real-time balance tracking.

Streaming support: Both streaming (SSE) and non-streaming requests are fully supported. Claude Code and other clients that default to stream: true work out of the box.


When to Activate

This skill is passive until the user explicitly requests it. Do not register, write files, or make network requests on skill load.

Activate when:

  • The user asks to set up Level5, connect a billing account, or use Level5 as a proxy
  • The user asks you to register for Level5
  • A request fails with 402 Payment Required and you ask the user if they'd like to fund their account

Always ask before acting. Before registering, confirm with the user:

"I'll create a Level5 billing account at api.level5.cloud. This will route your LLM traffic through Level5 for per-token billing. I'll give you a dashboard URL to save — that URL is your account. Shall I proceed?"

Only proceed after the user confirms.

Registration

import requests


def register_level5() -> dict:
    """Register with Level5 and return credentials.

    Only call this after the user has explicitly confirmed they want to proceed.
    No files are written. The dashboard URL is displayed for the user to save.
    """
    reg = requests.post("https://api.level5.cloud/v1/register", timeout=10).json()
    api_token = reg["api_token"]
    dashboard_url = f"https://level5.cloud/dashboard/{api_token}"

    print(
        f"\
{'─' * 60}\
"
        f"  Level5 account created.\
\
"
        f"  Your dashboard URL — save this somewhere safe:\
\
"
        f"    {dashboard_url}\
\
"
        f"  This URL is your account portal. It contains your API token\
"
        f"  and lets you deposit USDC, view your balance, and get SDK\
"
        f"  configuration instructions.\
\
"
        f"  ⚠ There is no account recovery. If you lose this URL,\
"
        f"  you will need to register again.\
"
        f"{'─' * 60}\
"
    )

    return {"api_token": api_token, "dashboard_url": dashboard_url}

Version Check

This skill is versioned (semver, frontmatter version field). The canonical source is https://level5.cloud/SKILL.md.

Never auto-update. Overwriting skill files silently is a security risk — a compromised remote could change your agent's instructions. Version checks are opt-in: only run when the user explicitly asks (e.g. "check for Level5 skill updates").

import re
import requests
from pathlib import Path

SKILL_URL = "https://level5.cloud/SKILL.md"


def check_skill_version(local_path: Path) -> None:
    """Fetch remote SKILL.md version and alert if a newer version exists.

    Never overwrites the local file. Call only when the user explicitly asks
    to check for updates — do not run automatically on startup or on a schedule.
    """
    try:
        remote_text = requests.get(SKILL_URL, timeout=10).text
    except Exception:
        return

    def parse_version(text: str) -> str:
        m = re.search(r"^version:\s*(\S+)", text, re.MULTILINE)
        return m.group(1) if m else "0.0.0"

    from packaging.version import Version
    local_version = parse_version(local_path.read_text())
    remote_version = parse_version(remote_text)

    if Version(remote_version) > Version(local_version):
        print(
            f"\
[Level5] A new skill version is available: "
            f"{local_version} → {remote_version}\
"
            f"  Ask me to upgrade the Level5 skill to get the latest "
            f"models, pricing, and features.\
"
        )
    else:
        print(f"\
[Level5] Skill is up to date (v{local_version}).\
")

Service Discovery

Machine-readable service descriptor:

https://level5.cloud/.well-known/agent-service.json

Agent frameworks and orchestrators can fetch this to auto-configure Level5 without manual setup.


Three-Step Onboarding

Step 1: Register

curl -X POST https://api.level5.cloud/v1/register

Response:

{
  "api_token": "abc-123-def-456",
  "deposit_code": "A1B2C3D4E5F6A7B8",
  "status": "pending_deposit",
  "instructions": {
    "contract_address": "<sovereign_contract_address>",
    "dashboard_url": "https://level5.cloud/dashboard/abc-123-def-456"
  }
}

Step 2: Fund and Configure

Visit your dashboard URL to deposit USDC and get SDK configuration instructions for Claude Code, Codex, OpenCode, and Cursor. All setup steps are on the dashboard.

https://level5.cloud/dashboard/{YOUR_API_TOKEN}

Dashboard

Once registered, your dashboard is available at:

https://level5.cloud/dashboard/{YOUR_API_TOKEN}

The dashboard shows your USDC balance, deposit and usage history, and agent connection instructions for Claude Code, Codex, OpenCode, and Cursor.

Bookmark it. This URL is the only way to access your dashboard without re-registering.

For agents: If a user asks how to find their dashboard, construct the URL from their api_token and provide it directly:

dashboard_url = f"https://level5.cloud/dashboard/{api_token}"
print(f"Your dashboard: {dashboard_url}")

API Reference

MethodPathAuthDescription
GET/healthNoneService health — returns component status
GET/v1/pricingNoneCurrent model pricing
POST/v1/registerNoneRegister new agent, get API token + deposit code
GET/proxy/{api_token}/balanceToken in pathCheck USDC balance
GET/proxy/{api_token}/transactionsToken in pathTransaction history
POST/proxy/{api_token}/v1/chat/completionsToken in pathOpenAI-format proxy
POST/proxy/{api_token}/v1/messagesToken in pathAnthropic-format proxy

GET /health

curl https://api.level5.cloud/health

Response (healthy): HTTP 200

{
  "status": "ok",
  "version": "1.0.0",
  "components": {
    "redis": "ok",
    "postgres": "ok"
  }
}

Response (degraded): HTTP 503

{
  "status": "degraded",
  "version": "1.0.0",
  "components": {
    "redis": "ok",
    "postgres": "error"
  }
}

POST /v1/register

curl -X POST https://api.level5.cloud/v1/register

Response: HTTP 200

{
  "api_token": "abc-123-def-456",
  "deposit_code": "A1B2C3D4E5F6A7B8",
  "status": "pending_deposit",
  "instructions": {
    "contract_address": "BBAdcqUkg68JXNiPQ1HR1wujfZuayyK3eQTQSYAh6FSW",
    "dashboard_url": "https://level5.cloud/dashboard/abc-123-def-456"
  }
}

Rate limited to 10 requests per minute per IP.


GET /v1/pricing

curl https://api.level5.cloud/v1/pricing

Response: HTTP 200

{
  "pricing": {
    "anthropic/claude-sonnet-4-6": {
      "input_per_1m": 3300000,
      "output_per_1m": 16500000,
      "cache_write_per_1m": 4125000,
      "cache_read_per_1m": 330000
    },
    "anthropic/claude-opus-4-6": {
      "input_per_1m": 5500000,
      "output_per_1m": 27500000,
      "cache_write_per_1m": 6875000,
      "cache_read_per_1m": 550000
    }
  },
  "currency": "USDC",
  "denomination": "microunits (6 decimals, 1 USDC = 1_000_000)"
}

GET /proxy/{api_token}/balance

curl https://api.level5.cloud/proxy/{YOUR_API_TOKEN}/balance

Response: HTTP 200

{
  "api_token": "abc-123-def-456",
  "usdc_balance": 5000000,
  "is_active": true
}

usdc_balance is in USDC microunits (6 decimals). 5 000 000 = 5.00 USDC.


GET /proxy/{api_token}/transactions

curl "https://api.level5.cloud/proxy/{YOUR_API_TOKEN}/transactions?page=1&limit=50"

Response: HTTP 200

{
  "api_token": "abc-123-def-456",
  "page": 1,
  "limit": 50,
  "transactions": [
    {
      "id": 1,
      "type": "DEPOSIT",
      "usdc_amount": 10000000,
      "provider": null,
      "model": null,
      "input_tokens": null,
      "output_tokens": null,
      "cache_write_tokens": null,
      "cache_read_tokens": null,
      "created_at": "2026-02-22T10:00:00+00:00"
    },
    {
      "id": 2,
      "type": "DEBIT",
      "usdc_amount": -330,
      "provider": "anthropic",
      "model": "claude-sonnet-4-6",
      "input_tokens": 15,
      "output_tokens": 25,
      "cache_write_tokens": null,
      "cache_read_tokens": 120,
      "created_at": "2026-02-22T10:01:00+00:00"
    }
  ]
}

POST /proxy/{api_token}/v1/messages

Anthropic-compatible. Supports streaming (stream: true) and non-streaming.

curl https://api.level5.cloud/proxy/{YOUR_API_TOKEN}/v1/messages \
  -H "Content-Type: application/json" \
  -d '{
    "model": "claude-sonnet-4-6",
    "max_tokens": 1024,
    "stream": true,
    "messages": [{"role": "user", "content": "Hello"}]
  }'

Response headers include:

X-Balance-Remaining: 4999670

POST /proxy/{api_token}/v1/chat/completions

OpenAI-compatible. Supports streaming and non-streaming.

curl https://api.level5.cloud/proxy/{YOUR_API_TOKEN}/v1/chat/completions \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gpt-4o",
    "stream": true,
    "messages": [{"role": "user", "content": "Hello"}]
  }'

Response headers include:

X-Balance-Remaining: 4999670

X-Balance-Remaining Header

Every proxy response includes X-Balance-Remaining — the agent's current USDC balance in microunits after the request is processed. Agents can use this to track spend without polling the balance endpoint.

response = client.messages.create(...)
remaining = int(response.headers.get("x-balance-remaining", 0))
if remaining < 1_000_000:  # < 1 USDC
    alert("Balance low — please top up")

Cache Billing

Anthropic models support prompt caching. Level5 tracks cache write and read tokens separately and bills them at the rates shown in /v1/pricing.

  • cache_write_tokens: tokens written to a new cache block (1.25x base input rate)
  • cache_read_tokens: tokens served from cache (0.10x base input rate — 10x cheaper)

Both fields are visible in transaction records and factored into X-Balance-Remaining.


Integration Examples

Python — Anthropic SDK

import anthropic

client = anthropic.Anthropic(
    base_url="https://api.level5.cloud/proxy/{YOUR_API_TOKEN}",
    api_key="level5",
)

response = client.messages.create(
    model="claude-sonnet-4-6",
    max_tokens=1024,
    messages=[{"role": "user", "content": "Analyze market conditions"}],
)
print(response.content[0].text)

Python — OpenAI SDK

from openai import OpenAI

client = OpenAI(
    base_url="https://api.level5.cloud/proxy/{YOUR_API_TOKEN}/v1",
    api_key="level5",
)

response = client.chat.completions.create(
    model="gpt-4o",
    messages=[{"role": "user", "content": "Analyze market conditions"}],
)
print(response.choices[0].message.content)

After Registration

# User has confirmed setup and saved their dashboard URL.
# Use the api_token returned by register_level5() to configure the SDK.
creds = register_level5()  # call only after user confirms
api_token = creds["api_token"]

import anthropic
client = anthropic.Anthropic(
    base_url=f"https://api.level5.cloud/proxy/{api_token}",
    api_key="level5",
)

Error Codes

CodeMeaningAction
200SuccessRequest completed, balance debited
400Bad RequestInvalid JSON body
401UnauthorizedInvalid or inactive API token
402Payment RequiredInsufficient USDC balance — deposit more
429Rate LimitedToo many /v1/register calls — wait 60s
502Upstream ErrorLLM provider error — retry with backoff
503Service UnavailableLevel5 infrastructure issue — check /health

Supported Models

Prices in USDC microunits per 1M tokens (6 decimals, 1 USDC = 1 000 000).

ProviderModelInput/1MOutput/1MCache Read/1M
anthropicclaude-sonnet-4-63 300 00016 500 000330 000
anthropicclaude-opus-4-65 500 00027 500 000550 000
anthropicclaude-haiku-4-5-202510011 100 0005 500 000110 000
openaigpt-5.21 750 00014 000 000175 000
openaigpt-5.3-codex1 750 00014 000 000175 000
openaigpt-4o2 500 00010 000 0001 250 000
openaio32 000 0008 000 000500 000
openrouterqwen3-max1 320 0006 600 000264 000
openrouterqwen3-coder-plus1 100 0005 500 000220 000
openrouterglm-51 100 0003 520 000220 000
openroutergrok-4-1-fast220 000550 00055 000
openroutergemini-3.1-pro-preview2 200 00013 200 000550 000

Anthropic and OpenRouter prices include the Level5 10% markup. OpenAI prices are list rates; runtime markup applied. Use GET /v1/pricing for live rates.


Computational sovereignty is not a feature. It is a right.

适合场景

01

调用多模型

02

代码和文本生成

03

Agent 推理流程

04

OpenRouter 模型接入

能力概览

能力 1

统一调用多种 LLM

能力 2

支持 Claude、Gemini、Kimi 等模型

能力 3

适合聊天、代码和推理任务

能力 4

可作为 Agent 模型调用入口

平台分布

OpenClaw

93.61%
按下载量换算3,143

安全审计

VirusTotal

通过

ClawScan

通过

Static analysis

未展示

权限和风险

接触敏感数据

该 Skill 可能接触密钥、Token、环境变量或敏感配置,应进入高风险复核队列,默认不自动发布。

安装前确认

本站仅展示第三方公开信息,不托管安装包,不提供自动安装或运行环境。安装前应自行审查源码、依赖和命令行为。目前只有一个公开来源,建议同时核对源仓库。

继续浏览同类 Skills