[codex] benchmark definition simplification refactor#480

Draft

MRiabov wants to merge 14 commits into

mainfrom

benchmark-definition-dedupe-refactor-start

Commits on Mar 29, 2026

(spec) Update benchmark constraints and role definitions in architecture and specs
MRiabov
committed
(refactor) Implement planner-authored solution estimates and derived caps
MRiabov
committed
(spec) Bulk update dataset seed artifacts and mock responses for cost/weight estimates
MRiabov
committed
(refactor) Update agent nodes and orchestration logic
MRiabov
committed
(devops) Update API tasks, benchmark routes, and integration runner
MRiabov
committed
(spec) Update prompts and reward configurations
MRiabov
committed
(test) Update integration tests for planner gates and node entry validation
MRiabov
committed
(spec) Bulk update dataset seed manifests and revisions (Round 3)
MRiabov
committed
(refactor) Update COTS runtime and workbench configuration
MRiabov
committed
(spec) Bulk update dataset seed artifacts and manifests
MRiabov
committed
(devops) Update dataset routes
MRiabov
committed
(test) Update integration tests for agent workflows and reviewer evidence
MRiabov
committed
(refactor) Refine benchmark assembly motion validation
MRiabov
committed
(spec) Plan for benchmark definition simplification refactor
MRiabov
committed