optimize-codebaselisted

# Optimize Codebase ## Purpose and authority Select one evidence-backed leverage point for a named workflow and seal one `optimization-handoff.md` for `plan-change`. Inspect without editing the target repository. Never implement, patch, prescribe file-level changes, write a test blueprint, or route directly to `implement-plan`. Keep drafts and measurements outside the installed skill, target repository, and final output directory. Ecosystem documentation may validate a locally selected mechanism; it never selects the mechanism. ## Start Resolve `skill-root` to this directory and pass absolute draft and output paths: ```bash python /absolute/skill-root/scripts/cli.py --repo-root /absolute/repo \ --input draft=/absolute/draft.md --input output_dir=/absolute/output \ --input scope=targeted --format json run ``` Run the returned `next_command.argv` with its returned `cwd`. Read only `required_reads`, write only `allowed_writes`, and stop on every `blocking_reason`. ## Workflow Follow [Optimization Contract](references/optimization-contract.md). Measure or bound the baseline, reconcile the requested workflow, compare independent candidates, and select one winner or a terminal state. Use [Optimization Techniques](references/optimization-techniques.md) only for the selected evidence branch. Set exactly one handoff state: `plan-ready`, `needs-evidence`, or `no-change`. A sweep accounts for every subsystem/pass pair and preserves resumable deferments. Paths and symbols a