research-overview-pipelinelisted

# research-overview-pipeline Goal: - Generate the final research overview from the top-ranked hypotheses. Inputs: - `research_plan/RESEARCH_PLAN.json` - top-ranked `hypotheses/<id>/HYPOTHESIS.json` artifacts with completed review payloads - optional `meta/INSIGHTS_FROM_REVIEWS.json` Outputs: - `meta/RESEARCH_OVERVIEW.json` - updated `state/PIPELINE_STATE.json` - updated `state/CURRENT_STAGE.json` Context Loading: - Read `research_plan/RESEARCH_PLAN.json`. - Read `packages/agent_contracts/pipeline_runtime.py` before updating `state/PIPELINE_STATE.json` or `state/CURRENT_STAGE.json`. - Read `state/EVOLUTION_STATE.json` before writing the overview. - Read `literature/queries/*/EVIDENCE_BUNDLE.json` when literature artifacts exist and record whether any `retrieval_metadata.status` is `partial` or `blocked`. - Read `state/PROXIMITY_STATUS.json` when it exists and record whether proximity was `succeeded` or degraded to a skipped/failed fallback state. - Use `research_goal` as the synthesis anchor. - Use `preferences` and `constraints` to frame which research directions are worth elevating. - Read the current top-ranked hypothesis artifacts together with their embedded or adjacent review results. - If `meta/INSIGHTS_FROM_REVIEWS.json` exists, use it to highlight recurring critique patterns and unresolved risks across the run. Execution Prompt Contract: - System Intent: - You are synthesizing the strongest current hypotheses into a final research overview. - Required Reas