Skip to main content
Glama
portel-dev

NCP - Natural Context Provider

by portel-dev
automation-comparison.md17.1 kB
# Code-Mode + Scheduler: Power Comparison ## Visual Token Cost Comparison ### Traditional Approach: find + run ``` ┌─────────────────────────────────────────────────────────────────┐ │ EVERY SINGLE EXECUTION │ │ │ │ Step 1: AI Discovery │ │ ┌──────────────────────────────────────────┐ │ │ │ find("check system health") │ → 2000 tokens │ │ │ Returns: system-health-check tool │ │ │ └──────────────────────────────────────────┘ │ │ ↓ │ │ Step 2: AI Execution │ │ ┌──────────────────────────────────────────┐ │ │ │ run("system-health-check", {}) │ → 500 tokens │ │ │ Returns: health metrics │ │ │ └──────────────────────────────────────────┘ │ │ ↓ │ │ Total per execution: 2500 tokens │ │ │ │ Daily (every minute = 1440 executions): │ │ 2500 × 1440 = 3,600,000 tokens │ │ │ │ 💸 Monthly Cost: ~$450 │ └─────────────────────────────────────────────────────────────────┘ ``` ### Code-Mode + Scheduler ``` ┌─────────────────────────────────────────────────────────────────┐ │ ONE-TIME SETUP │ │ │ │ ┌──────────────────────────────────────────┐ │ │ │ schedule.create({ │ │ │ │ name: "health-check", │ │ │ │ schedule: "* * * * *", │ → 1500 tokens │ │ │ tool: "code:run", │ (one-time) │ │ │ parameters: { code: "..." } │ │ │ │ }) │ │ │ └──────────────────────────────────────────┘ │ │ ↓ │ │ AUTOMATED EXECUTION │ │ ┌──────────────────────────────────────────┐ │ │ │ [No AI calls - runs automatically] │ → 50 tokens │ │ │ Scheduler executes code directly │ per run │ │ └──────────────────────────────────────────┘ │ │ ↓ │ │ Daily (every minute = 1440 executions): │ │ 50 × 1440 = 72,000 tokens │ │ │ │ 💰 Monthly Cost: ~$9 │ │ 🎉 SAVINGS: $441/month (98% reduction!) │ └─────────────────────────────────────────────────────────────────┘ ``` ## Feature Comparison Matrix ``` ┌───────────────────────────────────┬─────────────────┬─────────────────────┐ │ Feature │ find + run │ Code-Mode+Scheduler │ ├───────────────────────────────────┼─────────────────┼─────────────────────┤ │ Token Cost (per execution) │ 2500 tokens │ 50 tokens │ │ Setup Complexity │ N/A │ One-time │ │ AI Calls per Execution │ 2 calls │ 0 calls │ │ Multi-step Workflows │ ❌ Complex │ ✅ Easy │ │ Conditional Logic │ ❌ Limited │ ✅ Full JS │ │ Error Handling │ ❌ Manual │ ✅ Built-in │ │ Node.js Ecosystem Access │ ❌ No │ ✅ Yes │ │ Chain Operations │ ❌ Hard │ ✅ Easy │ │ Monthly Cost (1min interval) │ ~$450 │ ~$9 │ │ Cost Savings │ Baseline │ 98% │ └───────────────────────────────────┴─────────────────┴─────────────────────┘ ``` ## Scaling Comparison ### Cost Growth as Frequency Increases ``` Execution Frequency: Every 1 minute (1440/day) Traditional (find + run): ████████████████████████████████████████████████ $450/mo Code-Mode + Scheduler: █ $9/mo Execution Frequency: Every 5 minutes (288/day) Traditional (find + run): ██████████ $90/mo Code-Mode + Scheduler: ░ $2/mo Execution Frequency: Every hour (24/day) Traditional (find + run): ██ $7.50/mo Code-Mode + Scheduler: ░ $0.15/mo Legend: █ = $10 ░ = $0.10 ``` ## Complexity vs. Cost ### Simple Task (Single Operation) ``` Traditional: ┌────────────────────────┐ │ find() → 2000 tokens │ │ run() → 500 tokens │ │ Total: 2500 tokens │ └────────────────────────┘ Code-Mode: ┌────────────────────────┐ │ Setup: 1500 tokens │ │ Run: 50 tokens │ │ Pays off after: 1 run │ └────────────────────────┘ ``` ### Complex Workflow (5 Operations) ``` Traditional: ┌────────────────────────────────────────┐ │ find() → 2000 tokens │ │ run() → 500 tokens │ │ find() → 2000 tokens │ │ run() → 500 tokens │ │ find() → 2000 tokens │ │ run() → 500 tokens │ │ find() → 2000 tokens │ │ run() → 500 tokens │ │ find() → 2000 tokens │ │ run() → 500 tokens │ │ Total: 12,500 tokens per execution │ └────────────────────────────────────────┘ Code-Mode: ┌────────────────────────────────────────┐ │ Setup: 2000 tokens (one-time) │ │ Run: 50 tokens │ │ All 5 operations in single schedule! │ │ Pays off after: 1 run │ │ Savings: 99.6% per execution │ └────────────────────────────────────────┘ ``` ## Real-World Scenario Analysis ### Scenario 1: E-Commerce Inventory Management ``` Requirements: - Check inventory every 2 hours (12 times/day) - Update prices based on demand - Alert on low stock - Generate daily summary Traditional Approach: ┌─────────────────────────────────────────────────────┐ │ Per Check (4 operations): │ │ find inventory → 2000t │ │ run check → 500t │ │ find pricing → 2000t │ │ run update → 500t │ │ find alerts → 2000t │ │ run alert → 500t │ │ find summary → 2000t │ │ run generate → 500t │ │ Total: 10,000 tokens │ │ │ │ Daily: 10,000 × 12 = 120,000 tokens │ │ Monthly Cost: ~$150 │ └─────────────────────────────────────────────────────┘ Code-Mode + Scheduler: ┌─────────────────────────────────────────────────────┐ │ Setup: Single schedule with all logic (2000t) │ │ Per execution: 50 tokens │ │ │ │ Daily: 50 × 12 = 600 tokens │ │ Monthly Cost: ~$0.75 │ │ │ │ 💰 Savings: $149.25/month (99.5%) │ └─────────────────────────────────────────────────────┘ ``` ### Scenario 2: DevOps Health Monitoring ``` Requirements: - Check all services every 10 minutes (144 times/day) - Auto-rollback on failure - Alert on anomalies - Generate hourly reports Traditional Approach: ┌─────────────────────────────────────────────────────┐ │ Per Check: │ │ Multiple find/run cycles │ │ Estimated: 8,000 tokens │ │ │ │ Daily: 8,000 × 144 = 1,152,000 tokens │ │ Monthly Cost: ~$1,440 │ └─────────────────────────────────────────────────────┘ Code-Mode + Scheduler: ┌─────────────────────────────────────────────────────┐ │ Setup: Comprehensive monitoring script (2500t) │ │ Per execution: 50 tokens │ │ │ │ Daily: 50 × 144 = 7,200 tokens │ │ Monthly Cost: ~$9 │ │ │ │ 💰 Savings: $1,431/month (99.4%) │ └─────────────────────────────────────────────────────┘ ``` ## Capability Comparison ### What You CAN'T Do with Traditional find + run ``` ❌ Chain operations without multiple AI calls ❌ Use npm packages or Node.js built-ins ❌ Implement complex conditional logic ❌ Handle errors and retry automatically ❌ Access file system directly ❌ Make HTTP requests with custom logic ❌ Process data with custom algorithms ❌ Dynamic workflow adjustments ``` ### What You CAN Do with Code-Mode + Scheduler ``` ✅ Chain unlimited operations in one task ✅ Use entire Node.js ecosystem ✅ Implement any conditional logic ✅ Built-in error handling & retries ✅ Full file system access ✅ Custom HTTP clients & APIs ✅ Any data processing algorithm ✅ Self-adjusting workflows ✅ Multi-step ETL pipelines ✅ Complex business logic ✅ Integration with any service ✅ Custom reporting & analytics ``` ## ROI Timeline ``` Cost Comparison Over Time (1 minute execution frequency) Month 1: Traditional: $450 Code-Mode: $9 Savings: $441 Month 3: Traditional: $1,350 Code-Mode: $27 Savings: $1,323 Month 6: Traditional: $2,700 Code-Mode: $54 Savings: $2,646 Year 1: Traditional: $5,400 Code-Mode: $108 Savings: $5,292 🎉 Over 1 year with just ONE automation: ██████████████████████████████████████████████ $5,292 saved ``` ## When to Use Each Approach ### Use Traditional (find + run) When: ``` ✓ One-time operations ✓ Interactive AI decision-making required ✓ Exploratory tasks ✓ Simple tool discovery ✓ No scheduling needed ``` ### Use Code-Mode + Scheduler When: ``` ✓ Repeated executions (scheduled) ✓ Multi-step workflows ✓ Complex business logic ✓ Data pipelines ✓ System monitoring ✓ Automated reporting ✓ DevOps automation ✓ API integrations ✓ Cost optimization is priority ``` ## Migration Path ### Step 1: Identify Repetitive Tasks ``` Current workflow: AI: find("check logs") AI: run(log-checker) AI: find("send alert") AI: run(email-sender) ↓ This happens 288 times/day ↓ Cost: $90/month ``` ### Step 2: Consolidate into Code ```javascript // Single scheduled code block const logs = await checkLogs(); if (logs.hasErrors) { await sendAlert(logs.errors); } return { checked: logs.count, alerted: logs.hasErrors }; ``` ### Step 3: Schedule It ```javascript await schedule.create({ name: "log-monitoring", schedule: "*/5 * * * *", tool: "code:run", parameters: { code: "..." } }); ↓ Now runs automatically ↓ Cost: $2/month ↓ Savings: $88/month ``` ## Summary ``` ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┓ ┃ AUTOMATION POWERHOUSE ┃ ┃ ┃ ┃ Code-Mode + Scheduler = 98% Cost Reduction ┃ ┃ ┃ ┃ ✨ One-time setup instead of repeated AI calls ┃ ┃ 🚀 Execute any Node.js code on any schedule ┃ ┃ 💰 Massive token savings (50 vs 2500 tokens/execution) ┃ ┃ 🎯 Complex workflows in single schedule ┃ ┃ 📊 $441/month savings per automation ┃ ┃ ┃ ┃ The future of AI automation: Schedule intelligence, ┃ ┃ not just tools. ┃ ┗━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┛ ``` --- **Ready to start?** 1. 📖 Read: `examples/automation-powerhouse.md` 2. 🎯 Demo: `node examples/automation-demo.cjs` 3. 🧪 Test: `node tests/manual/test-automation-powerhouse.cjs` 4. 🚀 Create: `ncp run schedule:create`

Latest Blog Posts

MCP directory API

We provide all the information about MCP servers via our MCP API.

curl -X GET 'https://glama.ai/api/mcp/v1/servers/portel-dev/ncp'

If you have feedback or need assistance with the MCP directory API, please join our Discord server