Self-Improving Agent
一个面向 Automation 场景的 Agent 技能。原始说明:Captures learnings, errors, and corrections to enable continuous improvement. Use when: (1) A command or operation fails unexpectedly, (2) User corrects Clau...
一个面向 Automation 场景的 Agent 技能。原始说明:Create and run multi-agent AI workflows on AIProx. Chain agents into scheduled pipelines. Pay per execution in sats.
name: aiprox-workflows
description: Create and run multi-agent AI workflows on AIProx. Chain agents into scheduled pipelines. Pay per execution in sats.
acceptLicenseTerms: true
metadata:
clawdbot:
emoji: "⚡"
homepage: https://aiprox.dev/workflows
requires:
env:
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.
create_workflowCreate a named pipeline with ordered agent steps.
Create a workflow "daily-crypto-brief" that searches Bitcoin news, analyzes sentiment, and emails a summary @daily
Step chaining: use $step1.result, $step2.result, etc. to pass outputs forward.
Available capabilities:
| Capability | What it does |
|---|---|
| web-search | Real-time web search via search-bot |
| sentiment-analysis | Sentiment and tone analysis via sentiment-bot |
| scraping | Web scraping and article extraction via data-spider |
| data-analysis | Data processing and text analysis via doc-miner |
| translation | Multilingual translation via polyglot |
| vision | Image/screenshot analysis via vision-bot |
| code-execution | Code audit and security review via code-auditor |
| email | Send email notifications via email-bot |
| market-data | Prediction market signals via market-oracle |
| token-analysis | Solana token safety and rug detection via isitarug |
run_workflowTrigger any workflow by its ID.
Run workflow wf_abc123
list_workflowsShow all workflows for the current spend token.
List my workflows
get_run_historyShow past execution results, sats spent, and step-by-step output.
Show run history for wf_abc123
delete_workflowDelete a workflow (cancels scheduled runs).
run_templateRun a pre-built template by name in one shot.
Run the polymarket-signals template and email me at user@example.com
| Template name | Pipeline | Cost |
|---|---|---|
| news-digest | search-bot → sentiment-bot → email-bot | ~150 sats/run |
| token-scanner | data-spider → isitarug → email-bot | ~120 sats/run |
| competitive-intel | search-bot → doc-miner → sentiment-bot → email-bot | ~200 sats/run |
| multilingual-content | data-spider → doc-miner → polyglot | ~180 sats/run |
| site-audit | vision-bot → code-auditor → doc-miner | ~220 sats/run |
| polymarket-signals | market-oracle → sentiment-bot → email-bot | ~160 sats/run |
| api-security-audit | data-spider → code-auditor → pdf-bot → email-bot | ~300 sats/run |
Set AIPROX_SPEND_TOKEN in your MCP server config. Get a spend token at lightningprox.com.
{
"mcpServers": {
"aiprox-workflows": {
"command": "npx",
"args": ["aiprox-workflows-mcp"],
"env": {
"AIPROX_SPEND_TOKEN": "lnpx_your_token_here"
}
}
}
}
| Shorthand | When |
|---|---|
| @hourly | Every hour |
| @daily | Every day at midnight |
| @weekly | Every Sunday |
| 0 9 * * 1-5 | 9am Monday–Friday |
50–220 sats per workflow execution depending on agents used. No monthly plan. Deducted from your Lightning spend token balance.
When a user asks to automate something:
Cost transparency:
sats_spent after each runCheck templates first:
If the task matches a template, use run_template — it's faster and pre-configured.
For one-off orchestrate requests (not scheduled workflows), add callback_url to get a job_id back immediately and receive the result via webhook when done. Poll status at GET https://aiprox.dev/api/async/:job_id. Works with Zapier, Make, and n8n.
npm install aiprox-workflowsnpx aiprox-workflows-mcp