downstream-forecasting

Downstream Forecasting

Trace the cascade effects of a decision through its downstream dependents, with severity-gated termination and amplification detection. Produces a structured forecast showing where the decision propagates, at what severity, and where cascades terminate. Five phases: impact surface enumeration, cascade tracing, constraint check, interaction detection, forecast summary.

When to Use

• A decision has been made and you need to understand its downstream impact
• Tracing second-order effects that first-order thinking might miss
• Assessing severity of change propagation across systems or components
• After validating a claim via adversarial-cycle (Validate-and-Forecast pattern)
• After surfacing fragile assumptions (Assumption-and-Forecast pattern)

When NOT to Use

• The decision hasn't been validated yet (use adversarial-cycle first)
• You need to compare alternatives (use tradeoff-analysis)
• You need to surface hidden assumptions (use assumption-surfacing)
• The scope boundary is undefined - DF requires explicit scope to terminate

Execution

This protocol runs as the downstream-forecasting agent. The agent follows the full phase procedure autonomously, using the sequentialthinking MCP tool for phases marked (Sequential-thinking required).

• Protocol selection and composition: use /reason
• Domain research delegation: agent spawns domain-researcher as needed
• Phase validation: enforced by SubagentStop hook
• Without sequential-thinking: agent tags output validation_tier: Fallback

Input

• decision (required): The decision being forecasted, stated specifically and falsifiably
• scope_boundary (required): Systems, components, or domains in play; dependents outside are not traced
• decision_context (optional): Broader context from which the decision emerged
• severity_threshold (optional): Minimum severity to include in cascade tracing

Output

• impact_surface: Direct dependents with dependency mechanisms and amplification flags
• cascade_map: Per-dependent cascade paths with severity at each hop
• constraint_violations: Per-dependent constraint violations
• convergence_points: Components with converged impacts and re-evaluated severities
• boundary_exceeded: Cascade paths that reached the scope boundary
• forecast_summary: Human-readable per-dependent narrative
• validation_tier: Full | Fallback

Composition

• Validate-and-Forecast (V-F): Receives final_claim from Adversarial Cycle as decision input. Modified claims use the revised form.
• Assumption-and-Forecast (A-F): Receives with-basis assumptions from Assumption Surfacing via routing gate. without-basis assumptions route to coverage gap passthrough.
• T-V-F pipeline: Receives validated claim from the T-V stage.
• Use /reason for full composition guidance, pattern selection, and pipeline specifications.