A production-ready CLI for executing multi-stage AI workflows.
Cost tracking. Beautiful terminal UI. Zero config headaches.
No signup. No tracking. Just works.
$ ez-ai-pipeline run --pipeline prompt-optimizer --input "Write a blog post"
Pipeline: prompt-optimizer v1.0.0
4 stages configured
Stage 1/4: Analyzing input...
● claude-haiku-4 ($0.0012)
✓ Extracted 3 key themes
Stage 2/4: Generating structure...
● claude-sonnet-4 ($0.0089)
✓ Created 5-section outline
Stage 3/4: Enhancing prompt...
● claude-sonnet-4 ($0.0156)
✓ Added context and constraints
Stage 4/4: Validating output...
● claude-haiku-4 ($0.0008)
✓ Quality score: 94/100
Pipeline complete!
Duration: 12.3s | Total cost: $0.0265 | Tokens: 4,821
Output saved to: outputs/result-prompt-optimizer-2024-01-15.jsonBuilt for developers and power users who want AI workflows that actually work.
Chain AI calls together. Pass data between stages. Build complex workflows without the complexity.
Know exactly what you're spending. Per-stage costs, token counts, and totals. No surprise bills.
Progress bars, spinners, formatted output. Because staring at raw JSON is nobody's idea of fun.
Template variables that flow between stages. Reference outputs, chain results, build dynamic prompts.
Skip stages. Run conditionally. Exit early when done. Your pipelines, your rules.
Full control over execution. Made a mistake? Cancel and fix it. Need a break? Pause and come back.
Built-in web viewer for browsing outputs. Run ez-ai-pipeline web and see your results in the browser.
Ships with ready-to-use pipelines. Prompt optimizer, pipeline generator. Start using immediately.
If it takes more than 3 clicks, we screwed up.
npm install -g @ez-corp/ez-ai-pipeline
ez-ai-pipeline run -p prompt-optimizer -i "Your prompt"
ez-ai-pipeline web
No YAML nightmares. No config hell. Just code you can actually read.
import type { PipelineConfig } from "ez-ai-pipeline";
const config: PipelineConfig = {
pipeline: {
id: "my-pipeline",
name: "My Awesome Pipeline",
version: "1.0.0",
},
stages: [
{
id: "analyze",
name: "Analyze Input",
type: "analyze",
model: {
providerID: "anthropic",
modelID: "claude-haiku-4",
tier: "small", // Cost-efficient for analysis
},
prompt: {
template: `Analyze this: {{input}}`,
variables: [
{ name: "input", source: "input" }
],
},
output: { format: "json" },
},
{
id: "enhance",
name: "Enhance Results",
type: "enhance",
model: {
providerID: "anthropic",
modelID: "claude-sonnet-4",
tier: "medium", // Better quality for output
},
prompt: {
template: `Improve this: {{analysis}}`,
variables: [
{ name: "analysis", source: "previousStage" }
],
},
},
],
};
export default config;Anthropic, OpenAI, Google. Mix and match per stage.
Free forever. Open source. No credit card. No signup.
npm install -g @ez-corp/ez-ai-pipeline
export ANTHROPIC_API_KEY="your-key"
ez-ai-pipeline run -p prompt-optimizer -i "Your prompt"
ez-ai-pipeline web