executing-planslisted

# Executing plans The plan already did the thinking. Execute it exactly, prove each step with a test you watched fail and then pass, and leave a record that survives context compaction. **Required:** read test-driven-development before Task 1. It governs every step. ## Setup 1. Work on an isolated branch or worktree (git-workflow). Never implement on `main`/`master` without explicit consent. 2. Read the plan and its spec once. **The spec is the authority**; conflicts inside the plan resolve against it. 3. **Ledger:** create `docs/plans/<plan-name>.progress.md` (or the project's equivalent), first line `# Ledger — plan: <path>`. If it already exists and names this plan, tasks marked complete are **done**. Resume at the first incomplete one, and trust the ledger and `git log` over your memory. 4. **Pre-flight:** for every task that consumes another task's output, compare the Interfaces blocks. Record conflicts and your rulings in the ledger. 5. Run the verification commands once to record the baseline. Existing failures are noted in the ledger, not silently inherited. ## Per task 1. Re-read the task text (not your memory of it). Note `BASE=$(git rev-parse HEAD)`. 2. Work the steps in order. For every command, compare the output with the plan's `Expected`: - **Matches:** next step. - **The code is wrong:** use systematic-debugging. Never patch the symptom to match. - **The plan is wrong:** choose the smallest change that satisfies the spec, log `Ruling: <what> — <