Use when superpowers executing-plans is the correct fallback for an approved plan, especially when subagent dispatch is unavailable, tasks are tightly coupled, or the user explicitly wants single-thread execution.
This companion augments executing-plans. Core principle: execution mode must
be chosen before autonomy begins. executing-plans is the only fallback path
after subagent-first routing fails or is ruled out.
When to Use
The plan is approved
executing-plans is already relevant or subagent-first routing has been ruled out
The task is tightly coupled, the user explicitly wants single-thread execution, or a subagent-first attempt cannot actually dispatch in the current environment
Default finish policy should be verify -> commit -> push -> PR -> clean up
Chatty mid-flight confirmations are undesirable
The plan may live under a module directory instead of a flat
docs/superpowers/plans/ root
Do not use this when the current session can safely split work through
subagent-driven-development and dispatch can actually happen.
Rules
Related Skills
Before any substantial implementation, check whether subagent-first routing
applies.
If subagents are available and the plan can split by repo, module, or task,
route to superpowers:subagent-driven-development first; then let its
enhance handle autonomy.
If that subagent-first path cannot actually dispatch because of environment
limits, permission models, tool rules, or runtime failures,
executing-plans becomes the correct fallback and should take over
immediately.
If the task is tightly coupled or the user explicitly wants single-thread
execution, remain on executing-plans.
Once executing-plans is confirmed as the correct path, continue
automatically through implementation and normal finish steps.
Do not pause for routine "should I continue?" questions.
Interrupt only for auth problems, remote failures, real blockers, or
destructive risk.