Cost Time Optimizer

Cost Time Optimizer

Goal

Minimize execution cost and runtime without reducing required quality, policy compliance, or reproducibility.

Inputs

• workflow-spec.json
• execution-plan.json (if available)
• Existing artifacts in run directory
• Request constraints (deadline, depth, mandatory outputs)

Produce

Write optimization-plan.json with:

• baseline_plan_summary
• pruning_actions
• parallelization_actions
• reuse_actions
• prompt_budget_policy
• expected_savings (time and token estimate)
• risk_controls

Optimization policy

• Prune optional skills that do not affect user-stated outcomes.
• Reuse validated existing artifacts when skip conditions permit.
• Parallelize independent optional steps.
• Downscope overlong narrative artifacts when rubric/spec coverage is unchanged.
• Preserve all required quality gates.

Guardrails

• Never remove mandatory artifacts or required skills.
• Never bypass validation tools.
• If optimization increases semantic risk, downgrade confidence and require manual confirmation.

Rules

• Prefer deterministic, auditable reductions over aggressive heuristics.
• Every optimization action must include expected gain and rollback condition.