Stats
Actions
Tags
From polis
Execute an approved plan task by task via isolated subagents with fresh context. Enforces TDD, atomic commits, and two-phase (compliance + quality) review per task. Keeps the orchestrator under 40% and auto-pauses at WARNING.
How this command is triggered — by the user, by Claude, or both
Slash command
/polis:exec [feature name]The summary Claude sees in its command listing — used to decide when to auto-load this command
# /polis:exec
Run the approved plan. Runs the executing-plans skill (skills/executing-plans)
and dispatches via skills/subagent-dispatch.
## What it does
1. Load the latest approved `.claude/polis/plans/plan-<feature>-v<n>.md`. If
there's no approved plan, point to `/polis:plan`.
2. For each task in dependency order:
- Dispatch to a subagent with a fresh window: the task, the relevant spec
section, and minimal context — not the chat history or whole codebase.
- Subagent executes TDD (RED → GREEN → REFACTOR) and makes the atomic commit
(`[polis] T<n>: <what>`).
- Run...Run the approved plan. Runs the executing-plans skill (skills/executing-plans) and dispatches via skills/subagent-dispatch.
.claude/polis/plans/plan-<feature>-v<n>.md. If
there's no approved plan, point to /polis:plan.[polis] T<n>: <what>)./polis:pause-work) and resume with fresh context.npx claudepluginhub ilry-polis/polis --plugin polis/execExecutes a command inside a provisioned sandbox pod (defaults to interactive Claude session). Validates the bound claim and resolves the pod before exec.