Context Data Flow

examples/11-context-flow.halo

How results accumulate in context.data.* automatically WHY: Each pipeline stage's output is available to later stages

Open in the playground →

// ============================================================
// STEP 11: Context Data Flow
// ============================================================
// LEARN: How results accumulate in context.data.* automatically
// WHY:  Each pipeline stage's output is available to later stages
// ============================================================

agent Analyzer [openai/gpt-4o, temp: 0.5]

pipeline AnalyzeText(text: String) -> Result:
  text
    // Stage 1: Generate sentiment (stored to context.data.Analyzer.prompt)
    |> Analyzer.prompt("Analyze sentiment of: {{text}}")
    
    // Stage 2: Extract keywords (uses result from stage 1)
    // Access previous result via context.data
    |> Analyzer.prompt("Extract 3 keywords from previous analysis")
    
    // Stage 3: Generate summary using both previous results
    |> Analyzer.prompt("Write a one-line summary of the analysis")
    |> System.Print()

// NOTES:
// - context.data.Agent.Method holds the LAST result from that agent+method
// - For multiple results from same agent, use different methods or agents
// - Context is available to all subsequent stages in the same pipeline