AIProx Workflows

AIProx Workflows is a multi-agent pipeline engine. Chain AI agents into named workflows, schedule them to run on a cron, and get results delivered by email or webhook. Pay per execution in sats — no subscriptions, no monthly fees.

When to Use

- Building automated pipelines that chain multiple AI agents
Scheduling recurring tasks (daily news digest, competitive intel, market signals)
Running one-off multi-step pipelines on demand
Any workflow where one agent's output feeds into the next

MCP Tools

`create_workflow`

Create a named pipeline with ordered agent steps.

CODEBLOCK0

Step chaining: use $step1.result, $step2.result, etc. to pass outputs forward.

Available capabilities:

Capability	What it does
INLINECODE3	Real-time web search via search-bot
INLINECODE4

`run_workflow`

Trigger any workflow by its ID.

CODEBLOCK1

`list_workflows`

Show all workflows for the current spend token.

CODEBLOCK2

`get_run_history`

Show past execution results, sats spent, and step-by-step output.

CODEBLOCK3

`delete_workflow`

Delete a workflow (cancels scheduled runs).

`run_template`

Run a pre-built template by name in one shot.

CODEBLOCK4

Pre-Built Templates

Template name	Pipeline	Cost
INLINECODE18	search-bot → sentiment-bot → email-bot	~150 sats/run
INLINECODE19

Authentication

Set AIPROX_SPEND_TOKEN in your MCP server config. Get a spend token at lightningprox.com.

CODEBLOCK5

Scheduling

Shorthand	When
INLINECODE25	Every hour
INLINECODE26

Pricing

50–220 sats per workflow execution depending on agents used. No monthly plan. Deducted from your Lightning spend token balance.

Behavior Guidelines

When a user asks to automate something:

1. Identify which capabilities are needed (search → analyze → notify is the common pattern)
Suggest a template if one matches, otherwise offer to create a custom workflow
Confirm the schedule and email/webhook if they want recurring runs
After creating, offer to run it immediately

Cost transparency:

- Mention estimated sats cost before creating/running
Report sats_spent after each run

Check templates first:
If the task matches a template, use run_template — it's faster and pre-configured.

AIProx 工作流

AIProx 工作流是一个多智能体管道引擎。将AI智能体串联成命名工作流，按cron计划调度执行，并通过电子邮件或webhook获取结果。按执行次数付费（以聪为单位）——无订阅，无月费。

适用场景

- 构建串联多个AI智能体的自动化管道
调度重复性任务（每日新闻摘要、竞争情报、市场信号）
按需运行一次性多步骤管道
任何需要将一个智能体输出传递给下一个智能体的工作流

MCP 工具

create_workflow

创建包含有序智能体步骤的命名管道。

创建一个名为daily-crypto-brief的工作流，搜索比特币新闻，分析情感，并每天发送摘要邮件

步骤串联：使用 $step1.result、$step2.result 等将输出向前传递。

可用能力：

能力	功能说明
web-search	通过search-bot进行实时网络搜索
sentiment-analysis

run_workflow

通过工作流ID触发任何工作流。

运行工作流 wf_abc123

list_workflows

显示当前消费令牌的所有工作流。

列出我的工作流

getrunhistory

显示过去的执行结果、消耗的聪数以及逐步输出。

显示 wf_abc123 的运行历史

delete_workflow

删除工作流（取消计划运行）。

run_template

一键按名称运行预构建模板。

运行 polymarket-signals 模板，并将结果发送至 user@example.com

预构建模板

模板名称	管道	费用
news-digest	search-bot → sentiment-bot → email-bot	约150聪/次
token-scanner

身份验证

在MCP服务器配置中设置 AIPROXSPENDTOKEN。在 lightningprox.com 获取消费令牌。

json
{
mcpServers: {
aiprox-workflows: {
command: npx,
args: [aiprox-workflows-mcp],
env: {
AIPROXSPENDTOKEN: lnpxyourtoken_here
}
}
}
}

调度

简写	执行时间
@hourly	每小时
@daily

定价

每次工作流执行50–220聪，具体取决于使用的智能体。无月度计划。从您的闪电消费令牌余额中扣除。

行为指南

当用户要求自动化某项任务时：

1. 确定所需的能力（搜索→分析→通知是常见模式）
如有匹配模板则推荐，否则提供创建自定义工作流
确认调度计划和电子邮件/webhook（如需重复运行）
创建后，提供立即运行选项

成本透明：

- 在创建/运行前告知预估聪数成本
每次运行后报告 sats_spent

优先检查模板：
如果任务匹配模板，使用 run_template——更快且预配置。

链接

- 仪表盘：aiprox.dev/workflows
模板：aiprox.dev/templates
注册表：aiprox.dev/registry.html
npm SDK：npm install aiprox-workflows
MCP服务器：npx aiprox-workflows-mcp

aiprox-workflows AIProx工作流