executing-plans

1# Executing Plans
2 
3Execute the plan yourself, task by task, in this session: no implementer
4subagent per task, no reviewer per task. One fresh-context review of the
5whole branch at the end.
6 
7**Why inline:** Subagent-driven development pays for a fresh implementer
8and a fresh reviewer on every task, each re-reading the codebase from zero.
9Inline execution pays for one context (yours) plus one reviewer at the end.
10What it gives up is a fresh context per task and a second pair of eyes per
11task. This skill keeps what those two things bought, by other means: the
12brief is the spec, the ledger is your memory, TDD is the per-task gate, and
13the final reviewer is the second pair of eyes.
14 
15**Core principle:** The plan already did the thinking. Execute it exactly,
16prove each step with a test you watched fail and then pass, and leave a
17record that survives your own forgetting.
18 
19**Narration:** between tool calls, narrate at most one short line — the
20ledger and the tool results carry the record.
21 
22**Continuous execution:** Do not pause to check in with your human partner
23between tasks. They chose inline execution to spend less, not to answer
24"should I continue?" after every task. Execute all tasks from the plan
25without stopping.
26 
27**Rulings, not stalls.** Conflicts, ambiguities, plan defects — decide them.
28The spec is the binding authority, the plan is its argument, and your
29judgment settles what neither answers. Record every decision in the ledger
30…