Hypothesis and predicted-outcome generator using OpenAI

[CodeBuildder]

Context

OpenAI OpenAI model generates a structured hypothesis and predicted_outcome for the planned experiment. Parent: #29.

Acceptance Criteria

• OpenAI model prompt includes: target entity, fragility_score, recent findings, topology context
• Generated hypothesis follows the pattern: if [action] then [predicted_outcome]
• Prediction stored in Sentinel Operations Graph (SOG) as a prediction event validating against event.schema.json
• Prediction payload includes confidence (0–1) and predicted_outcome_type enum

Dependencies

Blocked by: none | Blocks: none

Notes

Model tier: OpenAI model.